Skip to content

Self-improving AI Employees for everyone

feat: add DeepSeek, Kimi, and Mistral LLM providers

trohitg90fbcebb9487
diff --git a/client/src/Dashboard.tsx b/client/src/Dashboard.tsx
index 8bb74613..e938168d 100644
--- a/client/src/Dashboard.tsx
+++ b/client/src/Dashboard.tsx
@@ -37,6 +37,7 @@ import { APIFY_NODE_TYPES } from './nodeDefinitions/apifyNodes';
import { GOOGLE_WORKSPACE_NODE_TYPES } from './nodeDefinitions/googleWorkspaceNodes';
import { PROXY_NODE_TYPES } from './nodeDefinitions/proxyNodes';
import { CRAWLEE_NODE_TYPES } from './nodeDefinitions/crawleeNodes';
+import { AI_CHAT_MODEL_TYPES } from './nodeDefinitions/aiModelNodes';
import ParameterPanel from './ParameterPanel';
import LocationParameterPanel from './components/LocationParameterPanel';
import { useAppStore } from './store/useAppStore';
@@ -86,7 +87,7 @@ const createNodeTypes = (): Record<string, React.ComponentType<any>> => {
// Trigger nodes - no input connections (start workflows)
if (TRIGGER_NODE_TYPES.includes(type)) {
types[type] = TriggerNode;
- } else if (type === 'openaiChatModel' || type === 'anthropicChatModel' || type === 'geminiChatModel' || type === 'openrouterChatModel' || type === 'groqChatModel' || type === 'cerebrasChatModel') {
+ } else if (AI_CHAT_MODEL_TYPES.includes(type)) {
// AI chat model nodes use square design
types[type] = SquareNode;
} else if (type === 'aiAgent' || type === 'chatAgent') {
diff --git a/client/src/assets/icons/llm/deepseek.svg b/client/src/assets/icons/llm/deepseek.svg
new file mode 100644
index 00000000..e64610fb
--- /dev/null
+++ b/client/src/assets/icons/llm/deepseek.svg
@@ -0,0 +1 @@
+<svg height="24" viewBox="0 0 24 24" width="24" xmlns="http://www.w3.org/2000/svg"><title>DeepSeek</title><path d="M23.748 4.482c-.254-.124-.364.113-.512.234-.051.039-.094.09-.137.136-.372.397-.806.657-1.373.626-.829-.046-1.537.214-2.163.848-.133-.782-.575-1.248-1.247-1.548-.352-.156-.708-.311-.955-.65-.172-.241-.219-.51-.305-.774-.055-.16-.11-.323-.293-.35-.2-.031-.278.136-.356.276-.313.572-.434 1.202-.422 1.84.027 1.436.633 2.58 1.838 3.393.137.093.172.187.129.323-.082.28-.18.552-.266.833-.055.179-.137.217-.329.14a5.526 5.526 0 01-1.736-1.18c-.857-.828-1.631-1.742-2.597-2.458a11.365 11.365 0 00-.689-.471c-.985-.957.13-1.743.388-1.836.27-.098.093-.432-.779-.428-.872.004-1.67.295-2.687.684a3.055 3.055 0 01-.465.137 9.597 9.597 0 00-2.883-.102c-1.885.21-3.39 1.102-4.497 2.623C.082 8.606-.231 10.684.152 12.85c.403 2.284 1.569 4.175 3.36 5.653 1.858 1.533 3.997 2.284 6.438 2.14 1.482-.085 3.133-.284 4.994-1.86.47.234.962.327 1.78.397.63.059 1.236-.03 1.705-.128.735-.156.684-.837.419-.961-2.155-1.004-1.682-.595-2.113-.926 1.096-1.296 2.746-2.642 3.392-7.003.05-.347.007-.565 0-.845-.004-.17.035-.237.23-.256a4.173 4.173 0 001.545-.475c1.396-.763 1.96-2.015 2.093-3.517.02-.23-.004-.467-.247-.588zM11.581 18c-2.089-1.642-3.102-2.183-3.52-2.16-.392.024-.321.471-.235.763.09.288.207.486.371.739.114.167.192.416-.113.603-.673.416-1.842-.14-1.897-.167-1.361-.802-2.5-1.86-3.301-3.307-.774-1.393-1.224-2.887-1.298-4.482-.02-.386.093-.522.477-.592a4.696 4.696 0 011.529-.039c2.132.312 3.946 1.265 5.468 2.774.868.86 1.525 1.887 2.202 2.891.72 1.066 1.494 2.082 2.48 2.914.348.292.625.514.891.677-.802.09-2.14.11-3.054-.614zm1-6.44a.306.306 0 01.415-.287.302.302 0 01.2.288.306.306 0 01-.31.307.303.303 0 01-.304-.308zm3.11 1.596c-.2.081-.399.151-.59.16a1.245 1.245 0 01-.798-.254c-.274-.23-.47-.358-.552-.758a1.73 1.73 0 01.016-.588c.07-.327-.008-.537-.239-.727-.187-.156-.426-.199-.688-.199a.559.559 0 01-.254-.078c-.11-.054-.2-.19-.114-.358.028-.054.16-.186.192-.21.356-.202.767-.136 1.146.016.352.144.618.408 1.001.782.391.451.462.576.685.914.176.265.336.537.445.848.067.195-.019.354-.25.452z" fill="#4D6BFE"/></svg>
diff --git a/client/src/assets/icons/llm/index.ts b/client/src/assets/icons/llm/index.ts
new file mode 100644
index 00000000..7cb06538
--- /dev/null
+++ b/client/src/assets/icons/llm/index.ts
@@ -0,0 +1,18 @@
+// LLM provider icons - extracted from @lobehub/icons
+// Following the same pattern as assets/icons/google/
+
+import deepseekSvg from './deepseek.svg?raw';
+import kimiSvg from './kimi.svg?raw';
+import mistralSvg from './mistral.svg?raw';
+
+// Convert SVG to data URI for use in node definitions
+const svgToDataUri = (svg: string): string => {
+ const encoded = encodeURIComponent(svg)
+ .replace(/'/g, '%27')
+ .replace(/"/g, '%22');
+ return `data:image/svg+xml,${encoded}`;
+};
+
+export const DEEPSEEK_ICON = svgToDataUri(deepseekSvg);
+export const KIMI_ICON = svgToDataUri(kimiSvg);
+export const MISTRAL_ICON = svgToDataUri(mistralSvg);
diff --git a/client/src/assets/icons/llm/kimi.svg b/client/src/assets/icons/llm/kimi.svg
new file mode 100644
index 00000000..d182fa4c
--- /dev/null
+++ b/client/src/assets/icons/llm/kimi.svg
@@ -0,0 +1 @@
+<svg fill="currentColor" fill-rule="evenodd" height="24" viewBox="0 0 24 24" width="24" xmlns="http://www.w3.org/2000/svg"><title>Kimi</title><path d="M19.738 5.776c.163-.209.306-.4.457-.585.07-.087.064-.153-.004-.244-.655-.861-.717-1.817-.34-2.787.283-.73.909-1.072 1.674-1.145.477-.045.945.004 1.379.236.57.305.902.77 1.01 1.412.086.512.07 1.012-.075 1.508-.257.878-.888 1.333-1.753 1.448-.718.096-1.446.108-2.17.157-.056.004-.113 0-.178 0z" fill="#027AFF"/><path d="M17.962 1.844h-4.326l-3.425 7.81H5.369V1.878H1.5V22h3.87v-8.477h6.824a3.025 3.025 0 002.743-1.75V22h3.87v-8.477a3.87 3.87 0 00-3.588-3.86v-.01h-2.125a3.94 3.94 0 002.323-2.12l2.545-5.689z"/></svg>
diff --git a/client/src/assets/icons/llm/mistral.svg b/client/src/assets/icons/llm/mistral.svg
new file mode 100644
index 00000000..0858a9a4
--- /dev/null
+++ b/client/src/assets/icons/llm/mistral.svg
@@ -0,0 +1 @@
+<svg height="24" viewBox="0 0 24 24" width="24" xmlns="http://www.w3.org/2000/svg"><title>Mistral</title><path d="M3.428 3.4h3.429v3.428H3.428V3.4zm13.714 0h3.43v3.428h-3.43V3.4z" fill="gold"/><path d="M3.428 6.828h6.857v3.429H3.429V6.828zm10.286 0h6.857v3.429h-6.857V6.828z" fill="#FFAF00"/><path d="M3.428 10.258h17.144v3.428H3.428v-3.428z" fill="#FF8205"/><path d="M3.428 13.686h3.429v3.428H3.428v-3.428zm6.858 0h3.429v3.428h-3.429v-3.428zm6.856 0h3.43v3.428h-3.43v-3.428z" fill="#FA500F"/><path d="M0 17.114h10.286v3.429H0v-3.429zm13.714 0H24v3.429H13.714v-3.429z" fill="#E10500"/></svg>
diff --git a/client/src/components/CredentialsModal.tsx b/client/src/components/CredentialsModal.tsx
index 492053a6..c2b4016b 100644
--- a/client/src/components/CredentialsModal.tsx
+++ b/client/src/components/CredentialsModal.tsx
@@ -24,6 +24,7 @@ import { useWhatsAppStatus, useAndroidStatus, useTwitterStatus, useGoogleStatus,
import { useWhatsApp } from '../hooks/useWhatsApp';
import {
OpenAIIcon, ClaudeIcon, GeminiIcon, GroqIcon, OpenRouterIcon, CerebrasIcon,
+ DeepSeekIcon, KimiIcon, MistralIcon, AI_PROVIDER_META,
} from './icons/AIProviderIcons';
import { ApifyIcons } from '../assets/icons/apify';
import { SearchIcons } from '../assets/icons/search';
@@ -123,12 +124,15 @@ const CATEGORIES: Category[] = [
key: 'ai',
label: 'AI Providers',
items: [
- { id: 'openai', name: 'OpenAI', placeholder: 'sk-...', color: '#10a37f', desc: 'GPT-4o, GPT-4, GPT-3.5', Icon: OpenAIIcon },
- { id: 'anthropic', name: 'Anthropic', placeholder: 'sk-ant-...', color: '#d97706', desc: 'Claude 3.5 Sonnet, Opus', Icon: ClaudeIcon },
- { id: 'gemini', name: 'Gemini', placeholder: 'AIza...', color: '#4285f4', desc: 'Gemini 2.0, 1.5 Pro/Flash', Icon: GeminiIcon },
- { id: 'groq', name: 'Groq', placeholder: 'gsk_...', color: '#F55036', desc: 'Llama, Mixtral - Ultra-fast', Icon: GroqIcon },
- { id: 'cerebras', name: 'Cerebras', placeholder: 'csk-...', color: '#FF6600', desc: 'Llama, Qwen - Ultra-fast', Icon: CerebrasIcon },
- { id: 'openrouter', name: 'OpenRouter', placeholder: 'sk-or-...', color: '#6366f1', desc: 'Unified API - 200+ models', Icon: OpenRouterIcon },
+ { id: 'openai', name: 'OpenAI', placeholder: 'sk-...', color: AI_PROVIDER_META.openai.color, desc: 'GPT-4o, GPT-4, GPT-3.5', Icon: OpenAIIcon },
+ { id: 'anthropic', name: 'Anthropic', placeholder: 'sk-ant-...', color: AI_PROVIDER_META.anthropic.color, desc: 'Claude 3.5 Sonnet, Opus', Icon: ClaudeIcon },
+ { id: 'gemini', name: 'Gemini', placeholder: 'AIza...', color: AI_PROVIDER_META.gemini.color, desc: 'Gemini 2.0, 1.5 Pro/Flash', Icon: GeminiIcon },
+ { id: 'groq', name: 'Groq', placeholder: 'gsk_...', color: AI_PROVIDER_META.groq.color, desc: 'Llama, Mixtral - Ultra-fast', Icon: GroqIcon },
+ { id: 'cerebras', name: 'Cerebras', placeholder: 'csk-...', color: AI_PROVIDER_META.cerebras.color, desc: 'Llama, Qwen - Ultra-fast', Icon: CerebrasIcon },
+ { id: 'openrouter', name: 'OpenRouter', placeholder: 'sk-or-...', color: AI_PROVIDER_META.openrouter.color, desc: 'Unified API - 200+ models', Icon: OpenRouterIcon },
+ { id: 'deepseek', name: 'DeepSeek', placeholder: 'sk-...', color: AI_PROVIDER_META.deepseek.color, desc: 'DeepSeek V3, Reasoner', Icon: DeepSeekIcon },
+ { id: 'kimi', name: 'Kimi', placeholder: 'sk-...', color: AI_PROVIDER_META.kimi.color, desc: 'Kimi K2.5 - Moonshot AI', Icon: KimiIcon },
+ { id: 'mistral', name: 'Mistral', placeholder: 'sk-...', color: AI_PROVIDER_META.mistral.color, desc: 'Mistral Large, Small, Codestral', Icon: MistralIcon },
],
},
{
diff --git a/client/src/components/ModelNode.tsx b/client/src/components/ModelNode.tsx
index fd08af1e..4a585981 100644
--- a/client/src/components/ModelNode.tsx
+++ b/client/src/components/ModelNode.tsx
@@ -41,6 +41,9 @@ const ModelNode: React.FC<NodeProps<NodeData>> = ({ id, type, data, isConnectabl
}
// Fallback: extract provider from node type
+ if (type?.includes('deepseek')) return 'deepseek';
+ if (type?.includes('kimi')) return 'kimi';
+ if (type?.includes('mistral')) return 'mistral';
if (type?.includes('cerebras')) return 'cerebras';
if (type?.includes('groq')) return 'groq';
if (type?.includes('openrouter')) return 'openrouter';
@@ -50,7 +53,6 @@ const ModelNode: React.FC<NodeProps<NodeData>> = ({ id, type, data, isConnectabl
if (type?.includes('azure')) return 'azure_openai';
if (type?.includes('cohere')) return 'cohere';
if (type?.includes('ollama')) return 'ollama';
- if (type?.includes('mistral')) return 'mistral';
return '';
}, [type, definition?.credentials]);
@@ -102,7 +104,6 @@ const ModelNode: React.FC<NodeProps<NodeData>> = ({ id, type, data, isConnectabl
if (type?.includes('azure')) return <span>☁️</span>;
if (type?.includes('cohere')) return <span>🌊</span>;
if (type?.includes('ollama')) return <span>🦙</span>;
- if (type?.includes('mistral')) return <span>🌪️</span>;
return <span>AI</span>;
};
diff --git a/client/src/components/ParameterRenderer.tsx b/client/src/components/ParameterRenderer.tsx
index e89c17e7..c001c847 100644
--- a/client/src/components/ParameterRenderer.tsx
+++ b/client/src/components/ParameterRenderer.tsx
@@ -13,19 +13,18 @@ import { useWebSocket } from '../contexts/WebSocketContext';
import { useApiKeys } from '../hooks/useApiKeys';
// Map node types to provider keys for AI model nodes
+import { AI_MODEL_PROVIDER_MAP } from '../nodeDefinitions/aiModelNodes';
+
+// Map node types to provider keys for AI model nodes
+// Uses the centralized map from aiModelNodes + legacy aliases
const NODE_TYPE_TO_PROVIDER: Record<string, string> = {
- 'openaiChatModel': 'openai',
- 'anthropicChatModel': 'anthropic',
+ ...AI_MODEL_PROVIDER_MAP,
+ // Legacy aliases for backward compatibility
'claudeChatModel': 'anthropic',
'googleChatModel': 'gemini',
- 'geminiChatModel': 'gemini',
'azureChatModel': 'azure_openai',
'cohereChatModel': 'cohere',
'ollamaChatModel': 'ollama',
- 'mistralChatModel': 'mistral',
- 'openrouterChatModel': 'openrouter',
- 'groqChatModel': 'groq',
- 'cerebrasChatModel': 'cerebras'
};
// Collection Renderer - n8n official style
diff --git a/client/src/components/SquareNode.tsx b/client/src/components/SquareNode.tsx
index 2c1be1e4..52f095ac 100644
--- a/client/src/components/SquareNode.tsx
+++ b/client/src/components/SquareNode.tsx
@@ -9,6 +9,7 @@ import { useWebSocket, useWhatsAppStatus } from '../contexts/WebSocketContext';
import { useApiKeys } from '../hooks/useApiKeys';
import { getAIProviderIcon } from './icons/AIProviderIcons';
import { PlayCircleFilled, ScheduleOutlined } from '@ant-design/icons';
+import { AI_MODEL_PROVIDER_MAP } from '../nodeDefinitions/aiModelNodes';
// Android service nodes that can connect to Android Toolkit as tools
const ANDROID_TOOL_CAPABLE_NODES = ANDROID_SERVICE_NODE_TYPES;
@@ -28,15 +29,8 @@ const WHATSAPP_NODE_TYPES = ['whatsappSend', 'whatsappReceive', 'whatsappDb'];
// Nodes that should not have output handles (input-only nodes)
const NO_OUTPUT_NODE_TYPES = ['console'];
-// AI Model node types with their provider IDs
-const AI_MODEL_NODE_TYPES: Record<string, string> = {
- 'openaiChatModel': 'openai',
- 'anthropicChatModel': 'anthropic',
- 'geminiChatModel': 'gemini',
- 'openrouterChatModel': 'openrouter',
- 'groqChatModel': 'groq',
- 'cerebrasChatModel': 'cerebras',
-};
+// AI Model node types with their provider IDs (derived from aiModelNodes registry)
+const AI_MODEL_NODE_TYPES = AI_MODEL_PROVIDER_MAP;
const SquareNode: React.FC<NodeProps<NodeData>> = ({ id, type, data, isConnectable, selected }) => {
const theme = useAppTheme();
diff --git a/client/src/components/icons/AIProviderIcons.tsx b/client/src/components/icons/AIProviderIcons.tsx
index 1bd26df2..8d475f4d 100644
--- a/client/src/components/icons/AIProviderIcons.tsx
+++ b/client/src/components/icons/AIProviderIcons.tsx
@@ -1,6 +1,7 @@
// AI Provider Icons - Using @lobehub/icons for official brand logos
import React from 'react';
-import { OpenAI, Claude, Gemini, Groq, OpenRouter, Cerebras } from '@lobehub/icons';
+import { OpenAI, Claude, Gemini, Groq, OpenRouter, Cerebras, DeepSeek, Kimi, Mistral } from '@lobehub/icons';
+import { dracula, solarized } from '../../styles/theme';
// Icon size constant for consistency
const ICON_SIZE = 28;
@@ -34,6 +35,18 @@ export const CerebrasIcon: React.FC<{ size?: number }> = ({ size = ICON_SIZE })
<Cerebras.Color size={size} />
);
+export const DeepSeekIcon: React.FC<{ size?: number }> = ({ size = ICON_SIZE }) => (
+ <DeepSeek.Color size={size} />
+);
+
+export const KimiIcon: React.FC<{ size?: number }> = ({ size = ICON_SIZE }) => (
+ <Kimi.Color size={size} />
+);
+
+export const MistralIcon: React.FC<{ size?: number }> = ({ size = ICON_SIZE }) => (
+ <Mistral.Color size={size} />
+);
+
// Map provider IDs to their icon components
export const AI_PROVIDER_ICONS: Record<string, React.FC<{ size?: number }>> = {
openai: OpenAIIcon,
@@ -42,16 +55,23 @@ export const AI_PROVIDER_ICONS: Record<string, React.FC<{ size?: number }>> = {
groq: GroqIcon,
openrouter: OpenRouterIcon,
cerebras: CerebrasIcon,
+ deepseek: DeepSeekIcon,
+ kimi: KimiIcon,
+ mistral: MistralIcon,
};
-// Centralized provider metadata (icon, brand color, display label)
+// Centralized provider metadata (icon, theme color, display label)
+// Colors use dracula/solarized palette for consistency with the rest of the UI
export const AI_PROVIDER_META: Record<string, { Icon: React.FC<{ size?: number }>; color: string; label: string }> = {
- openai: { Icon: OpenAIIcon, color: '#10a37f', label: 'OpenAI' },
- anthropic: { Icon: ClaudeIcon, color: '#d97706', label: 'Anthropic' },
- gemini: { Icon: GeminiIcon, color: '#4285f4', label: 'Gemini' },
- groq: { Icon: GroqIcon, color: '#F55036', label: 'Groq' },
- cerebras: { Icon: CerebrasIcon, color: '#FF6600', label: 'Cerebras' },
- openrouter: { Icon: OpenRouterIcon, color: '#6366f1', label: 'OpenRouter' },
+ openai: { Icon: OpenAIIcon, color: dracula.green, label: 'OpenAI' },
+ anthropic: { Icon: ClaudeIcon, color: dracula.orange, label: 'Anthropic' },
+ gemini: { Icon: GeminiIcon, color: solarized.blue, label: 'Gemini' },
+ groq: { Icon: GroqIcon, color: dracula.red, label: 'Groq' },
+ cerebras: { Icon: CerebrasIcon, color: dracula.orange, label: 'Cerebras' },
+ openrouter: { Icon: OpenRouterIcon, color: solarized.violet, label: 'OpenRouter' },
+ deepseek: { Icon: DeepSeekIcon, color: dracula.cyan, label: 'DeepSeek' },
+ kimi: { Icon: KimiIcon, color: dracula.purple, label: 'Kimi' },
+ mistral: { Icon: MistralIcon, color: dracula.pink, label: 'Mistral' },
};
// Get icon component by provider ID
diff --git a/client/src/components/ui/ComponentPalette.tsx b/client/src/components/ui/ComponentPalette.tsx
index 8a9268c0..051d1716 100644
--- a/client/src/components/ui/ComponentPalette.tsx
+++ b/client/src/components/ui/ComponentPalette.tsx
@@ -292,6 +292,8 @@ const ComponentPalette: React.FC<ComponentPaletteProps> = ({
dangerouslySetInnerHTML={{ __html: config.icon.svg }}
style={{ width: '16px', height: '16px', display: 'flex', alignItems: 'center', justifyContent: 'center' }}
/>
+ ) : typeof config.icon === 'string' && (config.icon.startsWith('data:') || config.icon.startsWith('http')) ? (
+ <img src={config.icon} alt="" style={{ width: '16px', height: '16px', objectFit: 'contain' }} />
) : (
config.icon
)}
diff --git a/client/src/factories/baseChatModelFactory.ts b/client/src/factories/baseChatModelFactory.ts
index 2b7f9995..8dd7ca7e 100644
--- a/client/src/factories/baseChatModelFactory.ts
+++ b/client/src/factories/baseChatModelFactory.ts
@@ -252,5 +252,24 @@ export const STANDARD_PARAMETERS = {
default: 'parsed',
description: 'How to return reasoning content (Groq)',
displayOptions: { show: { thinkingEnabled: [true] } }
- }
+ },
+
+ // Common request parameters (shared by all providers)
+ timeout: {
+ displayName: 'Timeout',
+ name: 'timeout',
+ type: 'number' as const,
+ default: 60000,
+ typeOptions: { minValue: 1000, maxValue: 180000 },
+ description: 'Timeout for the request in milliseconds'
+ },
+
+ maxRetries: {
+ displayName: 'Max Retries',
+ name: 'maxRetries',
+ type: 'number' as const,
+ default: 2,
+ typeOptions: { minValue: 0, maxValue: 5 },
+ description: 'Maximum number of retries'
+ },
};
\ No newline at end of file
diff --git a/client/src/nodeDefinitions/aiModelNodes.ts b/client/src/nodeDefinitions/aiModelNodes.ts
index df454037..f31fcf94 100644
--- a/client/src/nodeDefinitions/aiModelNodes.ts
+++ b/client/src/nodeDefinitions/aiModelNodes.ts
@@ -1,6 +1,8 @@
// AI Model Node Definitions - Standardized chat model configurations using base factory
import { INodeTypeDescription } from '../types/INodeProperties';
import { createBaseChatModel, ChatModelConfig, STANDARD_PARAMETERS } from '../factories/baseChatModelFactory';
+import { dracula } from '../styles/theme';
+import { DEEPSEEK_ICON, KIMI_ICON, MISTRAL_ICON } from '../assets/icons/llm';
// ============================================================================
// CHAT MODEL CONFIGURATIONS
@@ -13,7 +15,7 @@ const openaiConfig: ChatModelConfig = {
icon: '🤖',
color: '#00A67E',
description: 'OpenAI GPT models for chat completion and generation',
- models: [], // Models fetched dynamically via API
+ models: [],
parameters: [
STANDARD_PARAMETERS.frequencyPenalty,
{
@@ -27,14 +29,8 @@ const openaiConfig: ChatModelConfig = {
name: 'responseFormat',
type: 'options',
options: [
- {
- name: 'Text',
- value: 'text'
- },
- {
- name: 'JSON Object',
- value: 'json_object'
- }
+ { name: 'Text', value: 'text' },
+ { name: 'JSON Object', value: 'json_object' }
],
default: 'text',
description: 'Format of the response'
@@ -44,30 +40,9 @@ const openaiConfig: ChatModelConfig = {
...STANDARD_PARAMETERS.temperature,
description: 'Controls randomness (0-2). Note: O-series models (o1, o3, o4) only support temperature=1, which is set automatically.'
},
- {
- displayName: 'Timeout',
- name: 'timeout',
- type: 'number',
- default: 60000,
- typeOptions: {
- minValue: 1000,
- maxValue: 180000
- },
- description: 'Timeout for the request in milliseconds'
- },
- {
- displayName: 'Max Retries',
- name: 'maxRetries',
- type: 'number',
- default: 2,
- typeOptions: {
- minValue: 0,
- maxValue: 5
- },
- description: 'Maximum number of retries'
- },
+ STANDARD_PARAMETERS.timeout,
+ STANDARD_PARAMETERS.maxRetries,
STANDARD_PARAMETERS.topP,
- // Thinking/Reasoning for o-series models (o1, o3, o4)
STANDARD_PARAMETERS.thinkingEnabled,
STANDARD_PARAMETERS.reasoningEffort
]
@@ -80,11 +55,11 @@ const claudeConfig: ChatModelConfig = {
icon: '🧠',
color: '#FF6B35',
description: 'Anthropic Claude models for conversation and analysis',
- models: [], // Models fetched dynamically via API
+ models: [],
parameters: [
{
...STANDARD_PARAMETERS.maxTokens,
- default: 4096, // Higher default for Claude to accommodate thinking mode
+ default: 4096,
typeOptions: { minValue: 1, maxValue: 128000 },
description: 'Maximum tokens to generate. Must be greater than Thinking Budget when thinking is enabled.'
},
@@ -95,29 +70,8 @@ const claudeConfig: ChatModelConfig = {
},
STANDARD_PARAMETERS.topP,
STANDARD_PARAMETERS.topK,
- {
- displayName: 'Timeout',
- name: 'timeout',
- type: 'number',
- default: 60000,
- typeOptions: {
- minValue: 1000,
- maxValue: 180000
- },
- description: 'Timeout for the request in milliseconds'
- },
- {
- displayName: 'Max Retries',
- name: 'maxRetries',
- type: 'number',
- default: 2,
- typeOptions: {
- minValue: 0,
- maxValue: 5
- },
- description: 'Maximum number of retries'
- },
- // Extended thinking for Claude models
+ STANDARD_PARAMETERS.timeout,
+ STANDARD_PARAMETERS.maxRetries,
STANDARD_PARAMETERS.thinkingEnabled,
{
...STANDARD_PARAMETERS.thinkingBudget,
@@ -133,7 +87,7 @@ const geminiConfig: ChatModelConfig = {
icon: '⭐',
color: '#4285F4',
description: 'Google Gemini models for multimodal AI capabilities',
- models: [], // Models fetched dynamically via API
+ models: [],
parameters: [
{
...STANDARD_PARAMETERS.maxTokens,
@@ -146,55 +100,22 @@ const geminiConfig: ChatModelConfig = {
default: 0.9,
typeOptions: { minValue: 0, maxValue: 1, numberStepSize: 0.1 }
},
- {
- ...STANDARD_PARAMETERS.topP,
- default: 0.95
- },
+ { ...STANDARD_PARAMETERS.topP, default: 0.95 },
STANDARD_PARAMETERS.topK,
{
displayName: 'Safety Settings',
name: 'safetySettings',
type: 'options',
options: [
- {
- name: 'Default',
- value: 'default'
- },
- {
- name: 'Strict',
- value: 'strict'
- },
- {
- name: 'Permissive',
- value: 'permissive'
- }
+ { name: 'Default', value: 'default' },
+ { name: 'Strict', value: 'strict' },
+ { name: 'Permissive', value: 'permissive' }
],
default: 'default',
description: 'Content safety filtering level'
},
- {
- displayName: 'Timeout',
- name: 'timeout',
- type: 'number',
- default: 60000,
- typeOptions: {
- minValue: 1000,
- maxValue: 180000
- },
- description: 'Timeout for the request in milliseconds'
- },
- {
- displayName: 'Max Retries',
- name: 'maxRetries',
- type: 'number',
- default: 2,
- typeOptions: {
- minValue: 0,
- maxValue: 5
- },
- description: 'Maximum number of retries'
- },
- // Thinking for Gemini models - uses thinking_budget (token count)
+ STANDARD_PARAMETERS.timeout,
+ STANDARD_PARAMETERS.maxRetries,
STANDARD_PARAMETERS.thinkingEnabled,
{
...STANDARD_PARAMETERS.thinkingBudget,
@@ -210,35 +131,15 @@ const openrouterConfig: ChatModelConfig = {
icon: '🔀',
color: '#6366F1',
description: 'OpenRouter unified API - access OpenAI, Claude, Gemini, Llama, and more through one API',
- models: [], // Models fetched dynamically via API
+ models: [],
parameters: [
STANDARD_PARAMETERS.temperature,
STANDARD_PARAMETERS.maxTokens,
STANDARD_PARAMETERS.topP,
STANDARD_PARAMETERS.frequencyPenalty,
STANDARD_PARAMETERS.presencePenalty,
- {
- displayName: 'Timeout',
- name: 'timeout',
- type: 'number',
- default: 60000,
- typeOptions: {
- minValue: 1000,
- maxValue: 180000
- },
- description: 'Timeout for the request in milliseconds'
- },
- {
- displayName: 'Max Retries',
- name: 'maxRetries',
- type: 'number',
- default: 2,
- typeOptions: {
- minValue: 0,
- maxValue: 5
- },
- description: 'Maximum number of retries'
- }
+ STANDARD_PARAMETERS.timeout,
+ STANDARD_PARAMETERS.maxRetries,
]
};
@@ -249,34 +150,13 @@ const groqConfig: ChatModelConfig = {
icon: '⚡',
color: '#F55036',
description: 'Groq - Ultra-fast LLM inference with Llama, Mixtral, and Gemma models',
- models: [], // Models fetched dynamically via API
+ models: [],
parameters: [
STANDARD_PARAMETERS.temperature,
STANDARD_PARAMETERS.maxTokens,
STANDARD_PARAMETERS.topP,
- {
- displayName: 'Timeout',
- name: 'timeout',
- type: 'number',
- default: 60000,
- typeOptions: {
- minValue: 1000,
- maxValue: 180000
- },
- description: 'Timeout for the request in milliseconds'
- },
- {
- displayName: 'Max Retries',
- name: 'maxRetries',
- type: 'number',
- default: 2,
- typeOptions: {
- minValue: 0,
- maxValue: 5
- },
- description: 'Maximum number of retries'
- },
- // Reasoning for Groq models (Qwen3, QwQ)
+ STANDARD_PARAMETERS.timeout,
+ STANDARD_PARAMETERS.maxRetries,
STANDARD_PARAMETERS.thinkingEnabled,
STANDARD_PARAMETERS.reasoningFormat
]
@@ -287,38 +167,83 @@ const cerebrasConfig: ChatModelConfig = {
providerId: 'cerebras',
displayName: 'Cerebras',
icon: '🧬',
- color: '#FF6600',
+ color: dracula.orange,
description: 'Cerebras - Ultra-fast inference with Llama and Qwen models on custom AI hardware',
- models: [], // Models fetched dynamically via API
+ models: [],
parameters: [
STANDARD_PARAMETERS.temperature,
STANDARD_PARAMETERS.maxTokens,
STANDARD_PARAMETERS.topP,
+ STANDARD_PARAMETERS.timeout,
+ STANDARD_PARAMETERS.maxRetries,
+ STANDARD_PARAMETERS.thinkingEnabled,
+ STANDARD_PARAMETERS.thinkingBudget
+ ]
+};
+
+// DeepSeek Chat Model Configuration
+const deepseekConfig: ChatModelConfig = {
+ providerId: 'deepseek',
+ displayName: 'DeepSeek',
+ icon: DEEPSEEK_ICON,
+ color: dracula.cyan,
+ description: 'DeepSeek V3 models with Chain-of-Thought reasoning (deepseek-chat, deepseek-reasoner)',
+ models: [],
+ parameters: [
{
- displayName: 'Timeout',
- name: 'timeout',
- type: 'number',
- default: 60000,
- typeOptions: {
- minValue: 1000,
- maxValue: 180000
- },
- description: 'Timeout for the request in milliseconds'
+ ...STANDARD_PARAMETERS.temperature,
+ description: 'Controls randomness (0-2). Ignored by deepseek-reasoner.'
},
+ STANDARD_PARAMETERS.maxTokens,
+ STANDARD_PARAMETERS.topP,
+ STANDARD_PARAMETERS.frequencyPenalty,
+ STANDARD_PARAMETERS.presencePenalty,
+ STANDARD_PARAMETERS.timeout,
+ STANDARD_PARAMETERS.maxRetries,
+ STANDARD_PARAMETERS.thinkingEnabled,
+ ]
+};
+
+// Kimi (Moonshot AI) Chat Model Configuration
+const kimiConfig: ChatModelConfig = {
+ providerId: 'kimi',
+ displayName: 'Kimi',
+ icon: KIMI_ICON,
+ color: dracula.purple,
+ description: 'Kimi K2 models by Moonshot AI with 256K context and reasoning support',
+ models: [],
+ parameters: [
{
- displayName: 'Max Retries',
- name: 'maxRetries',
- type: 'number',
- default: 2,
- typeOptions: {
- minValue: 0,
- maxValue: 5
- },
- description: 'Maximum number of retries'
+ ...STANDARD_PARAMETERS.temperature,
+ typeOptions: { minValue: 0, maxValue: 1, numberStepSize: 0.1 },
+ description: 'Controls randomness (0-1). Fixed at 1.0 for thinking models.'
},
- // Thinking for Cerebras models (Qwen, Llama)
+ STANDARD_PARAMETERS.maxTokens,
+ STANDARD_PARAMETERS.topP,
+ STANDARD_PARAMETERS.timeout,
+ STANDARD_PARAMETERS.maxRetries,
STANDARD_PARAMETERS.thinkingEnabled,
- STANDARD_PARAMETERS.thinkingBudget
+ ]
+};
+
+// Mistral AI Chat Model Configuration
+const mistralConfig: ChatModelConfig = {
+ providerId: 'mistral',
+ displayName: 'Mistral',
+ icon: MISTRAL_ICON,
+ color: dracula.orange,
+ description: 'Mistral AI models for reasoning, coding, and multilingual tasks',
+ models: [],
+ parameters: [
+ {
+ ...STANDARD_PARAMETERS.temperature,
+ typeOptions: { minValue: 0, maxValue: 1, numberStepSize: 0.1 },
+ description: 'Controls randomness (0-1)'
+ },
+ STANDARD_PARAMETERS.maxTokens,
+ STANDARD_PARAMETERS.topP,
+ STANDARD_PARAMETERS.timeout,
+ STANDARD_PARAMETERS.maxRetries,
]
};
@@ -332,9 +257,20 @@ export const aiModelNodes: Record<string, INodeTypeDescription> = {
geminiChatModel: createBaseChatModel(geminiConfig),
openrouterChatModel: createBaseChatModel(openrouterConfig),
groqChatModel: createBaseChatModel(groqConfig),
- cerebrasChatModel: createBaseChatModel(cerebrasConfig)
+ cerebrasChatModel: createBaseChatModel(cerebrasConfig),
+ deepseekChatModel: createBaseChatModel(deepseekConfig),
+ kimiChatModel: createBaseChatModel(kimiConfig),
+ mistralChatModel: createBaseChatModel(mistralConfig),
};
-// Export configurations and factory for external use
-export { openaiConfig, claudeConfig, geminiConfig, openrouterConfig, groqConfig, cerebrasConfig, createBaseChatModel };
+// Node type names for external reference (Dashboard, SquareNode, etc.)
+export const AI_CHAT_MODEL_TYPES = Object.keys(aiModelNodes);
+
+// Map node type -> provider ID for icon/credential lookup
+const ALL_CONFIGS = [openaiConfig, claudeConfig, geminiConfig, openrouterConfig, groqConfig, cerebrasConfig, deepseekConfig, kimiConfig, mistralConfig];
+export const AI_MODEL_PROVIDER_MAP: Record<string, string> = Object.fromEntries(
+ ALL_CONFIGS.map(c => [`${c.providerId}ChatModel`, c.providerId])
+);
+// Export configurations and factory for external use
+export { openaiConfig, claudeConfig, geminiConfig, openrouterConfig, groqConfig, cerebrasConfig, deepseekConfig, kimiConfig, mistralConfig, createBaseChatModel };
diff --git a/client/src/services/executionService.ts b/client/src/services/executionService.ts
index 409573cc..28fde06b 100644
--- a/client/src/services/executionService.ts
+++ b/client/src/services/executionService.ts
@@ -11,6 +11,7 @@ import { APIFY_NODE_TYPES } from '../nodeDefinitions/apifyNodes';
import { SEARCH_NODE_TYPES } from '../nodeDefinitions/searchNodes';
import { TELEGRAM_NODE_TYPES } from '../nodeDefinitions/telegramNodes';
import { CRAWLEE_NODE_TYPES } from '../nodeDefinitions/crawleeNodes';
+import { AI_CHAT_MODEL_TYPES } from '../nodeDefinitions/aiModelNodes';
import { Node, Edge } from 'reactflow';
import { INodeExecutionData } from '../types/INodeProperties';
import { API_CONFIG } from '../config/api';
@@ -199,16 +200,7 @@ export class ExecutionService {
// Specialized Agent Nodes (use chatAgent backend handler)
...SPECIALIZED_AGENT_TYPES,
// Note: simpleMemory is passive - accessed by AI Agent automatically, not executed directly
- 'openaiChatModel',
- 'anthropicChatModel',
- 'geminiChatModel',
- 'openrouterChatModel',
- 'groqChatModel',
- 'cerebrasChatModel',
- 'azure_openaiChatModel', // Planned
- 'cohereChatModel', // Planned
- 'ollamaChatModel', // Planned
- 'mistralChatModel', // Planned
+ ...AI_CHAT_MODEL_TYPES,
// Google Maps Nodes
'gmaps_create',
'gmaps_locations',
diff --git a/server/config/llm_defaults.json b/server/config/llm_defaults.json
index b0dbcbd5..e8aac9e5 100644
--- a/server/config/llm_defaults.json
+++ b/server/config/llm_defaults.json
@@ -1,12 +1,16 @@
{
- "version": "2026.03",
- "last_updated": "2026-03-10",
- "_description": "3 latest + 2 popular models for OpenAI/Anthropic/Gemini, full support for Groq/Cerebras. Auto-enriched by model_registry.json (OpenRouter cache).",
+ "version": "2026.04",
+ "last_updated": "2026-04-03",
+ "_description": "LLM provider defaults. Connection config (base_url, models_endpoint) drives both native SDK and LangChain paths. Model specs auto-enriched by model_registry.json (OpenRouter cache).",
"providers": {
"openai": {
"default_model": "gpt-5.4",
"detection_patterns": ["gpt", "openai", "o1", "o3", "o4"],
+ "base_url": "https://api.openai.com/v1",
+ "models_endpoint": "https://api.openai.com/v1/models",
+ "api_key_param": "openai_api_key",
+ "max_tokens_param": "max_tokens",
"max_output_tokens": {
"gpt-5.4": 128000,
"gpt-5.3-chat": 16384,
@@ -31,6 +35,10 @@
"anthropic": {
"default_model": "claude-sonnet-4-6",
"detection_patterns": ["claude", "anthropic"],
+ "base_url": "https://api.anthropic.com/v1",
+ "models_endpoint": "https://api.anthropic.com/v1/models",
+ "api_key_param": "anthropic_api_key",
+ "max_tokens_param": "max_tokens",
"max_output_tokens": {
"claude-opus-4-6": 128000,
"claude-sonnet-4-6": 128000,
@@ -53,6 +61,10 @@
"gemini": {
"default_model": "gemini-2.5-flash",
"detection_patterns": ["gemini", "google"],
+ "base_url": "https://generativelanguage.googleapis.com/v1beta",
+ "models_endpoint": "https://generativelanguage.googleapis.com/v1beta/models",
+ "api_key_param": "google_api_key",
+ "max_tokens_param": "max_output_tokens",
"max_output_tokens": {
"gemini-3.1-pro-preview": 65536,
"gemini-3-flash-preview": 65536,
@@ -76,6 +88,11 @@
"openrouter": {
"default_model": "anthropic/claude-sonnet-4.6",
"detection_patterns": ["openrouter"],
+ "base_url": "https://openrouter.ai/api/v1",
+ "models_endpoint": "https://openrouter.ai/api/v1/models",
+ "api_key_param": "api_key",
+ "max_tokens_param": "max_tokens",
+ "extra_headers": {"HTTP-Referer": "http://localhost:3000", "X-Title": "MachinaOS"},
"max_output_tokens": {
"_default": 4096
},
@@ -87,14 +104,16 @@
"groq": {
"default_model": "llama-4-scout",
"detection_patterns": ["groq"],
+ "base_url": "https://api.groq.com/openai/v1",
+ "models_endpoint": "https://api.groq.com/openai/v1/models",
+ "api_key_param": "api_key",
+ "max_tokens_param": "max_tokens",
"max_output_tokens": {
"llama-4-scout": 8192,
"llama-4-maverick": 16384,
"llama-3.3-70b": 32768,
"llama-3.1-8b": 131072,
"qwen3-32b": 40960,
- "gpt-oss-120b": 65536,
- "gpt-oss-20b": 65536,
"_default": 8192
},
"context_length": {
@@ -103,8 +122,6 @@
"llama-3.3-70b": 131072,
"llama-3.1-8b": 131072,
"qwen3-32b": 131072,
- "gpt-oss-120b": 131072,
- "gpt-oss-20b": 131072,
"_default": 131072
},
"thinking_models": ["qwen3"],
@@ -112,23 +129,113 @@
"temperature_range": [0.0, 2.0]
},
"cerebras": {
- "default_model": "qwen-3-235b",
- "detection_patterns": ["cerebras", "llama3.1", "gpt-oss", "qwen-3-235b"],
+ "default_model": "llama3.1-8b",
+ "detection_patterns": ["cerebras", "llama3.1", "qwen-3-235b"],
+ "base_url": "https://api.cerebras.ai/v1",
+ "models_endpoint": "https://api.cerebras.ai/v1/models",
+ "api_key_param": "api_key",
+ "max_tokens_param": "max_tokens",
"max_output_tokens": {
"qwen-3-235b": 8192,
- "gpt-oss-120b": 8192,
"llama3.1-8b": 8192,
"_default": 8192
},
"context_length": {
"qwen-3-235b": 32768,
- "gpt-oss-120b": 131072,
"llama3.1-8b": 128000,
"_default": 128000
},
"thinking_models": ["qwen-3-235b"],
"thinking_type": "budget",
"temperature_range": [0.0, 1.5]
+ },
+ "xai": {
+ "default_model": "grok-3",
+ "detection_patterns": ["grok", "xai"],
+ "base_url": "https://api.x.ai/v1",
+ "models_endpoint": "https://api.x.ai/v1/models",
+ "api_key_param": "api_key",
+ "max_tokens_param": "max_tokens",
+ "max_output_tokens": {
+ "grok-3": 131072,
+ "_default": 8192
+ },
+ "context_length": {
+ "grok-3": 131072,
+ "_default": 131072
+ },
+ "temperature_range": [0.0, 2.0]
+ },
+ "deepseek": {
+ "default_model": "deepseek-chat",
+ "detection_patterns": ["deepseek"],
+ "base_url": "https://api.deepseek.com",
+ "models_endpoint": "https://api.deepseek.com/models",
+ "api_key_param": "api_key",
+ "max_tokens_param": "max_tokens",
+ "max_output_tokens": {
+ "deepseek-chat": 8192,
+ "deepseek-reasoner": 65536,
+ "_default": 8192
+ },
+ "context_length": {
+ "deepseek-chat": 131072,
+ "deepseek-reasoner": 131072,
+ "_default": 131072
+ },
+ "thinking_models": ["deepseek-reasoner", "deepseek-chat"],
+ "thinking_type": "enabled",
+ "temperature_range": [0.0, 2.0]
+ },
+ "kimi": {
+ "default_model": "kimi-k2.5",
+ "detection_patterns": ["kimi", "moonshot"],
+ "base_url": "https://api.moonshot.ai/v1",
+ "models_endpoint": "https://api.moonshot.ai/v1/models",
+ "api_key_param": "api_key",
+ "max_tokens_param": "max_tokens",
+ "max_output_tokens": {
+ "kimi-k2.5": 98304,
+ "kimi-k2-thinking": 98304,
+ "moonshot-v1-8k": 8192,
+ "moonshot-v1-32k": 32768,
+ "moonshot-v1-128k": 131072,
+ "_default": 8192
+ },
+ "context_length": {
+ "kimi-k2.5": 262144,
+ "kimi-k2-thinking": 262144,
+ "moonshot-v1-8k": 8192,
+ "moonshot-v1-32k": 32768,
+ "moonshot-v1-128k": 131072,
+ "_default": 262144
+ },
+ "thinking_models": ["kimi-k2.5", "kimi-k2-thinking"],
+ "thinking_type": "enabled",
+ "temperature_range": [0.0, 1.0]
+ },
+ "mistral": {
+ "default_model": "mistral-large-latest",
+ "detection_patterns": ["mistral", "codestral"],
+ "base_url": "https://api.mistral.ai/v1",
+ "models_endpoint": "https://api.mistral.ai/v1/models",
+ "api_key_param": "api_key",
+ "max_tokens_param": "max_tokens",
+ "max_output_tokens": {
+ "mistral-large-latest": 131072,
+ "mistral-small-latest": 32768,
+ "codestral-latest": 32768,
+ "open-mistral-nemo": 131072,
+ "_default": 8192
+ },
+ "context_length": {
+ "mistral-large-latest": 262144,
+ "mistral-small-latest": 32768,
+ "codestral-latest": 262144,
+ "open-mistral-nemo": 131072,
+ "_default": 131072
+ },
+ "temperature_range": [0.0, 1.0]
}
}
}
diff --git a/server/config/model_registry.json b/server/config/model_registry.json
index 2c12cc44..a70ba74f 100644
--- a/server/config/model_registry.json
+++ b/server/config/model_registry.json
@@ -1,8 +1,200 @@
{
- "_generated": "2026-03-30T06:51:06.462851+00:00",
+ "_generated": "2026-04-02T10:34:48.408712+00:00",
"_source": "openrouter",
- "_model_count": 345,
+ "_model_count": 350,
"models": {
+ "z-ai/glm-5v-turbo": {
+ "id": "z-ai/glm-5v-turbo",
+ "name": "Z.ai: GLM 5V Turbo",
+ "provider": "z-ai",
+ "local_id": "glm-5v-turbo",
+ "context_length": 202752,
+ "max_output_tokens": 131072,
+ "input_price_per_mtok": 1.2,
+ "output_price_per_mtok": 4.0,
+ "supports_thinking": false,
+ "thinking_type": "none",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "include_reasoning",
+ "max_tokens",
+ "reasoning",
+ "response_format",
+ "temperature",
+ "tool_choice",
+ "tools",
+ "top_p"
+ ]
+ },
+ "arcee-ai/trinity-large-thinking": {
+ "id": "arcee-ai/trinity-large-thinking",
+ "name": "Arcee AI: Trinity Large Thinking",
+ "provider": "arcee-ai",
+ "local_id": "trinity-large-thinking",
+ "context_length": 262144,
+ "max_output_tokens": 80000,
+ "input_price_per_mtok": 0.25,
+ "output_price_per_mtok": 0.9,
+ "supports_thinking": false,
+ "thinking_type": "none",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "include_reasoning",
+ "max_tokens",
+ "reasoning",
+ "temperature",
+ "tool_choice",
+ "tools",
+ "top_k",
+ "top_p"
+ ]
+ },
+ "xai/grok-4.20-multi-agent": {
+ "id": "x-ai/grok-4.20-multi-agent",
+ "name": "xAI: Grok 4.20 Multi-Agent",
+ "provider": "xai",
+ "local_id": "grok-4.20-multi-agent",
+ "context_length": 2000000,
+ "max_output_tokens": 0,
+ "input_price_per_mtok": 2.0,
+ "output_price_per_mtok": 6.0,
+ "supports_thinking": false,
+ "thinking_type": "none",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "include_reasoning",
+ "logprobs",
+ "max_tokens",
+ "reasoning",
+ "response_format",
+ "seed",
+ "structured_outputs",
+ "temperature",
+ "top_logprobs",
+ "top_p"
+ ]
+ },
+ "xai/grok-4.20": {
+ "id": "x-ai/grok-4.20",
+ "name": "xAI: Grok 4.20",
+ "provider": "xai",
+ "local_id": "grok-4.20",
+ "context_length": 2000000,
+ "max_output_tokens": 0,
+ "input_price_per_mtok": 2.0,
+ "output_price_per_mtok": 6.0,
+ "supports_thinking": false,
+ "thinking_type": "none",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "include_reasoning",
+ "logprobs",
+ "max_tokens",
+ "reasoning",
+ "response_format",
+ "seed",
+ "structured_outputs",
+ "temperature",
+ "tool_choice",
+ "tools",
+ "top_logprobs",
+ "top_p"
+ ]
+ },
+ "gemini/lyria-3-pro-preview": {
+ "id": "google/lyria-3-pro-preview",
+ "name": "Google: Lyria 3 Pro Preview",
+ "provider": "gemini",
+ "local_id": "lyria-3-pro-preview",
+ "context_length": 1048576,
+ "max_output_tokens": 65536,
+ "input_price_per_mtok": 0.0,
+ "output_price_per_mtok": 0.0,
+ "supports_thinking": false,
+ "thinking_type": "none",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "max_tokens",
+ "response_format",
+ "seed",
+ "temperature",
+ "top_p"
+ ]
+ },
+ "gemini/lyria-3-clip-preview": {
+ "id": "google/lyria-3-clip-preview",
+ "name": "Google: Lyria 3 Clip Preview",
+ "provider": "gemini",
+ "local_id": "lyria-3-clip-preview",
+ "context_length": 1048576,
+ "max_output_tokens": 65536,
+ "input_price_per_mtok": 0.0,
+ "output_price_per_mtok": 0.0,
+ "supports_thinking": false,
+ "thinking_type": "none",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "max_tokens",
+ "response_format",
+ "seed",
+ "temperature",
+ "top_p"
+ ]
+ },
+ "qwen/qwen3.6-plus-preview:free": {
+ "id": "qwen/qwen3.6-plus-preview:free",
+ "name": "Qwen: Qwen3.6 Plus Preview (free)",
+ "provider": "qwen",
+ "local_id": "qwen3.6-plus-preview:free",
+ "context_length": 1000000,
+ "max_output_tokens": 32000,
+ "input_price_per_mtok": 0.0,
+ "output_price_per_mtok": 0.0,
+ "supports_thinking": false,
+ "thinking_type": "none",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "include_reasoning",
+ "max_tokens",
+ "presence_penalty",
+ "reasoning",
+ "response_format",
+ "seed",
+ "structured_outputs",
+ "temperature",
+ "tool_choice",
+ "tools",
+ "top_p"
+ ]
+ },
"kwaipilot/kat-coder-pro-v2": {
"id": "kwaipilot/kat-coder-pro-v2",
"name": "Kwaipilot: KAT-Coder-Pro V2",
@@ -37,10 +229,10 @@
"top_p"
]
},
- "reka/reka-edge": {
- "id": "reka/reka-edge",
+ "rekaai/reka-edge": {
+ "id": "rekaai/reka-edge",
"name": "Reka Edge",
- "provider": "reka",
+ "provider": "rekaai",
"local_id": "reka-edge",
"context_length": 16384,
"max_output_tokens": 16384,
@@ -275,66 +467,6 @@
"top_p"
]
},
- "xai/grok-4.20-multi-agent-beta": {
- "id": "x-ai/grok-4.20-multi-agent-beta",
- "name": "xAI: Grok 4.20 Multi-Agent Beta",
- "provider": "xai",
- "local_id": "grok-4.20-multi-agent-beta",
- "context_length": 2000000,
- "max_output_tokens": 0,
- "input_price_per_mtok": 2.0,
- "output_price_per_mtok": 6.0,
- "supports_thinking": false,
- "thinking_type": "none",
- "temperature_range": [
- 0.0,
- 2.0
- ],
- "is_reasoning_model": false,
- "supported_parameters": [
- "include_reasoning",
- "logprobs",
- "max_tokens",
- "reasoning",
- "response_format",
- "seed",
- "structured_outputs",
- "temperature",
- "top_logprobs",
- "top_p"
- ]
- },
- "xai/grok-4.20-beta": {
- "id": "x-ai/grok-4.20-beta",
- "name": "xAI: Grok 4.20 Beta",
- "provider": "xai",
- "local_id": "grok-4.20-beta",
- "context_length": 2000000,
- "max_output_tokens": 0,
- "input_price_per_mtok": 2.0,
- "output_price_per_mtok": 6.0,
- "supports_thinking": false,
- "thinking_type": "none",
- "temperature_range": [
- 0.0,
- 2.0
- ],
- "is_reasoning_model": false,
- "supported_parameters": [
- "include_reasoning",
- "logprobs",
- "max_tokens",
- "reasoning",
- "response_format",
- "seed",
- "structured_outputs",
- "temperature",
- "tool_choice",
- "tools",
- "top_logprobs",
- "top_p"
- ]
- },
"nvidia/nemotron-3-super-120b-a12b:free": {
"id": "nvidia/nemotron-3-super-120b-a12b:free",
"name": "NVIDIA: Nemotron 3 Super (free)",
@@ -434,7 +566,7 @@
"provider": "qwen",
"local_id": "qwen3.5-9b",
"context_length": 256000,
- "max_output_tokens": 65536,
+ "max_output_tokens": 32768,
"input_price_per_mtok": 0.05,
"output_price_per_mtok": 0.15,
"supports_thinking": false,
@@ -1073,10 +1205,10 @@
"name": "MiniMax: MiniMax M2.5",
"provider": "minimax",
"local_id": "minimax-m2.5",
- "context_length": 196608,
- "max_output_tokens": 65536,
- "input_price_per_mtok": 0.19,
- "output_price_per_mtok": 1.15,
+ "context_length": 196600,
+ "max_output_tokens": 196600,
+ "input_price_per_mtok": 0.118,
+ "output_price_per_mtok": 1.25,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -1309,7 +1441,7 @@
"provider": "stepfun",
"local_id": "step-3.5-flash",
"context_length": 262144,
- "max_output_tokens": 0,
+ "max_output_tokens": 65536,
"input_price_per_mtok": 0.1,
"output_price_per_mtok": 0.3,
"supports_thinking": false,
@@ -1331,6 +1463,7 @@
"seed",
"stop",
"temperature",
+ "tool_choice",
"tools",
"top_k",
"top_p"
@@ -1368,9 +1501,9 @@
"provider": "moonshotai",
"local_id": "kimi-k2.5",
"context_length": 262144,
- "max_output_tokens": 65535,
- "input_price_per_mtok": 0.42,
- "output_price_per_mtok": 2.2,
+ "max_output_tokens": 262144,
+ "input_price_per_mtok": 0.3827,
+ "output_price_per_mtok": 1.9092,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -1561,6 +1694,8 @@
"stop",
"structured_outputs",
"temperature",
+ "tool_choice",
+ "tools",
"top_logprobs",
"top_p"
]
@@ -1592,6 +1727,8 @@
"stop",
"structured_outputs",
"temperature",
+ "tool_choice",
+ "tools",
"top_logprobs",
"top_p"
]
@@ -2861,40 +2998,6 @@
"tools"
]
},
- "kwaipilot/kat-coder-pro": {
- "id": "kwaipilot/kat-coder-pro",
- "name": "Kwaipilot: KAT-Coder-Pro V1",
- "provider": "kwaipilot",
- "local_id": "kat-coder-pro",
- "context_length": 256000,
- "max_output_tokens": 128000,
- "input_price_per_mtok": 0.207,
- "output_price_per_mtok": 0.828,
- "supports_thinking": false,
- "thinking_type": "none",
- "temperature_range": [
- 0.0,
- 2.0
- ],
- "is_reasoning_model": false,
- "supported_parameters": [
- "frequency_penalty",
- "logit_bias",
- "max_tokens",
- "min_p",
- "presence_penalty",
- "repetition_penalty",
- "response_format",
- "seed",
- "stop",
- "structured_outputs",
- "temperature",
- "tool_choice",
- "tools",
- "top_k",
- "top_p"
- ]
- },
"moonshotai/kimi-k2-thinking": {
"id": "moonshotai/kimi-k2-thinking",
"name": "MoonshotAI: Kimi K2 Thinking",
@@ -3627,6 +3730,7 @@
"is_reasoning_model": false,
"supported_parameters": [
"frequency_penalty",
+ "logit_bias",
"max_tokens",
"min_p",
"presence_penalty",
@@ -3828,7 +3932,7 @@
"provider": "gemini",
"local_id": "gemini-2.5-flash-lite-preview-09-2025",
"context_length": 1048576,
- "max_output_tokens": 65536,
+ "max_output_tokens": 65535,
"input_price_per_mtok": 0.1,
"output_price_per_mtok": 0.4,
"supports_thinking": true,
@@ -5490,19 +5594,15 @@
"is_reasoning_model": false,
"supported_parameters": [
"frequency_penalty",
- "logprobs",
"max_tokens",
"presence_penalty",
"repetition_penalty",
- "response_format",
"seed",
"stop",
- "structured_outputs",
"temperature",
"tool_choice",
"tools",
"top_k",
- "top_logprobs",
"top_p"
]
},
@@ -7375,6 +7475,35 @@
"web_search_options"
]
},
+ "rekaai/reka-flash-3": {
+ "id": "rekaai/reka-flash-3",
+ "name": "Reka Flash 3",
+ "provider": "rekaai",
+ "local_id": "reka-flash-3",
+ "context_length": 65536,
+ "max_output_tokens": 65536,
+ "input_price_per_mtok": 0.1,
+ "output_price_per_mtok": 0.2,
+ "supports_thinking": false,
+ "thinking_type": "none",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "frequency_penalty",
+ "include_reasoning",
+ "max_tokens",
+ "presence_penalty",
+ "reasoning",
+ "seed",
+ "stop",
+ "temperature",
+ "top_k",
+ "top_p"
+ ]
+ },
"gemini/gemma-3-27b-it:free": {
"id": "google/gemma-3-27b-it:free",
"name": "Google: Gemma 3 27B (free)",
@@ -8044,7 +8173,6 @@
"is_reasoning_model": false,
"supported_parameters": [
"frequency_penalty",
- "logit_bias",
"max_tokens",
"min_p",
"presence_penalty",
@@ -8053,8 +8181,6 @@
"seed",
"stop",
"temperature",
- "tool_choice",
- "tools",
"top_k",
"top_p"
]
@@ -8736,6 +8862,32 @@
"top_p"
]
},
+ "anthropic/claude-3.5-sonnet": {
+ "id": "anthropic/claude-3.5-sonnet",
+ "name": "Anthropic: Claude 3.5 Sonnet",
+ "provider": "anthropic",
+ "local_id": "claude-3.5-sonnet",
+ "context_length": 200000,
+ "max_output_tokens": 8192,
+ "input_price_per_mtok": 6.0,
+ "output_price_per_mtok": 30.0,
+ "supports_thinking": true,
+ "thinking_type": "budget",
+ "temperature_range": [
+ 0.0,
+ 1.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "max_tokens",
+ "stop",
+ "temperature",
+ "tool_choice",
+ "tools",
+ "top_k",
+ "top_p"
+ ]
+ },
"anthracite-org/magnum-v4-72b": {
"id": "anthracite-org/magnum-v4-72b",
"name": "Magnum v4 72B",
@@ -8770,32 +8922,6 @@
"top_p"
]
},
- "anthropic/claude-3.5-sonnet": {
- "id": "anthropic/claude-3.5-sonnet",
- "name": "Anthropic: Claude 3.5 Sonnet",
- "provider": "anthropic",
- "local_id": "claude-3.5-sonnet",
- "context_length": 200000,
- "max_output_tokens": 8192,
- "input_price_per_mtok": 6.0,
- "output_price_per_mtok": 30.0,
- "supports_thinking": true,
- "thinking_type": "budget",
- "temperature_range": [
- 0.0,
- 1.0
- ],
- "is_reasoning_model": false,
- "supported_parameters": [
- "max_tokens",
- "stop",
- "temperature",
- "tool_choice",
- "tools",
- "top_k",
- "top_p"
- ]
- },
"qwen/qwen-2.5-7b-instruct": {
"id": "qwen/qwen-2.5-7b-instruct",
"name": "Qwen: Qwen2.5 7B Instruct",
@@ -9432,11 +9558,11 @@
"top_p"
]
},
- "openai/gpt-4o-mini": {
- "id": "openai/gpt-4o-mini",
- "name": "OpenAI: GPT-4o-mini",
+ "openai/gpt-4o-mini-2024-07-18": {
+ "id": "openai/gpt-4o-mini-2024-07-18",
+ "name": "OpenAI: GPT-4o-mini (2024-07-18)",
"provider": "openai",
- "local_id": "gpt-4o-mini",
+ "local_id": "gpt-4o-mini-2024-07-18",
"context_length": 128000,
"max_output_tokens": 16384,
"input_price_per_mtok": 0.15,
@@ -9466,11 +9592,11 @@
"web_search_options"
]
},
- "openai/gpt-4o-mini-2024-07-18": {
- "id": "openai/gpt-4o-mini-2024-07-18",
- "name": "OpenAI: GPT-4o-mini (2024-07-18)",
+ "openai/gpt-4o-mini": {
+ "id": "openai/gpt-4o-mini",
+ "name": "OpenAI: GPT-4o-mini",
"provider": "openai",
- "local_id": "gpt-4o-mini-2024-07-18",
+ "local_id": "gpt-4o-mini",
"context_length": 128000,
"max_output_tokens": 16384,
"input_price_per_mtok": 0.15,
@@ -9613,15 +9739,15 @@
"top_p"
]
},
- "openai/gpt-4o": {
- "id": "openai/gpt-4o",
- "name": "OpenAI: GPT-4o",
+ "openai/gpt-4o-2024-05-13": {
+ "id": "openai/gpt-4o-2024-05-13",
+ "name": "OpenAI: GPT-4o (2024-05-13)",
"provider": "openai",
- "local_id": "gpt-4o",
+ "local_id": "gpt-4o-2024-05-13",
"context_length": 128000,
- "max_output_tokens": 16384,
- "input_price_per_mtok": 2.5,
- "output_price_per_mtok": 10.0,
+ "max_output_tokens": 4096,
+ "input_price_per_mtok": 5.0,
+ "output_price_per_mtok": 15.0,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -9647,15 +9773,15 @@
"web_search_options"
]
},
- "openai/gpt-4o:extended": {
- "id": "openai/gpt-4o:extended",
- "name": "OpenAI: GPT-4o (extended)",
+ "openai/gpt-4o": {
+ "id": "openai/gpt-4o",
+ "name": "OpenAI: GPT-4o",
"provider": "openai",
- "local_id": "gpt-4o:extended",
+ "local_id": "gpt-4o",
"context_length": 128000,
- "max_output_tokens": 64000,
- "input_price_per_mtok": 6.0,
- "output_price_per_mtok": 18.0,
+ "max_output_tokens": 16384,
+ "input_price_per_mtok": 2.5,
+ "output_price_per_mtok": 10.0,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -9681,15 +9807,15 @@
"web_search_options"
]
},
- "openai/gpt-4o-2024-05-13": {
- "id": "openai/gpt-4o-2024-05-13",
- "name": "OpenAI: GPT-4o (2024-05-13)",
+ "openai/gpt-4o:extended": {
+ "id": "openai/gpt-4o:extended",
+ "name": "OpenAI: GPT-4o (extended)",
"provider": "openai",
- "local_id": "gpt-4o-2024-05-13",
+ "local_id": "gpt-4o:extended",
"context_length": 128000,
- "max_output_tokens": 4096,
- "input_price_per_mtok": 5.0,
- "output_price_per_mtok": 15.0,
+ "max_output_tokens": 64000,
+ "input_price_per_mtok": 6.0,
+ "output_price_per_mtok": 18.0,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -10326,11 +10452,11 @@
"top_p"
]
},
- "openai/gpt-4": {
- "id": "openai/gpt-4",
- "name": "OpenAI: GPT-4",
+ "openai/gpt-4-0314": {
+ "id": "openai/gpt-4-0314",
+ "name": "OpenAI: GPT-4 (older v0314)",
"provider": "openai",
- "local_id": "gpt-4",
+ "local_id": "gpt-4-0314",
"context_length": 8191,
"max_output_tokens": 4096,
"input_price_per_mtok": 30.0,
@@ -10359,11 +10485,11 @@
"top_p"
]
},
- "openai/gpt-4-0314": {
- "id": "openai/gpt-4-0314",
- "name": "OpenAI: GPT-4 (older v0314)",
+ "openai/gpt-4": {
+ "id": "openai/gpt-4",
+ "name": "OpenAI: GPT-4",
"provider": "openai",
- "local_id": "gpt-4-0314",
+ "local_id": "gpt-4",
"context_length": 8191,
"max_output_tokens": 4096,
"input_price_per_mtok": 30.0,
diff --git a/server/config/pricing.json b/server/config/pricing.json
index 3a2da963..2f6c9b2c 100644
--- a/server/config/pricing.json
+++ b/server/config/pricing.json
@@ -33,18 +33,35 @@
"llama-3.3-70b": {"input": 0.10, "output": 0.10},
"llama-3.1-8b": {"input": 0.05, "output": 0.08},
"qwen3-32b": {"input": 0.29, "output": 0.59},
- "gpt-oss-120b": {"input": 0.15, "output": 0.60},
- "gpt-oss-20b": {"input": 0.10, "output": 0.10},
"_default": {"input": 0.15, "output": 0.60}
},
"cerebras": {
"qwen-3-235b": {"input": 0.20, "output": 0.20},
- "gpt-oss-120b": {"input": 0.10, "output": 0.10},
"llama3.1-8b": {"input": 0.10, "output": 0.10},
"_default": {"input": 0.10, "output": 0.10}
},
"openrouter": {
"_default": {"input": 1.00, "output": 5.00}
+ },
+ "deepseek": {
+ "deepseek-chat": {"input": 0.28, "output": 0.42, "cache_read": 0.028},
+ "deepseek-reasoner": {"input": 0.55, "output": 2.19, "cache_read": 0.14},
+ "_default": {"input": 0.28, "output": 0.42}
+ },
+ "kimi": {
+ "kimi-k2.5": {"input": 0.60, "output": 2.50, "cache_read": 0.15},
+ "kimi-k2-thinking": {"input": 0.60, "output": 2.50, "cache_read": 0.15},
+ "moonshot-v1-8k": {"input": 0.20, "output": 2.00},
+ "moonshot-v1-32k": {"input": 1.00, "output": 3.00},
+ "moonshot-v1-128k": {"input": 2.00, "output": 5.00},
+ "_default": {"input": 0.60, "output": 2.50}
+ },
+ "mistral": {
+ "mistral-large-latest": {"input": 2.00, "output": 6.00},
+ "mistral-small-latest": {"input": 0.20, "output": 0.60},
+ "codestral-latest": {"input": 0.30, "output": 0.90},
+ "open-mistral-nemo": {"input": 0.15, "output": 0.15},
+ "_default": {"input": 0.20, "output": 0.60}
}
},
diff --git a/server/constants.py b/server/constants.py
index 8776f6e8..ddeab495 100644
--- a/server/constants.py
+++ b/server/constants.py
@@ -17,6 +17,9 @@ AI_CHAT_MODEL_TYPES: FrozenSet[str] = frozenset([
'openrouterChatModel',
'groqChatModel',
'cerebrasChatModel',
+ 'deepseekChatModel',
+ 'kimiChatModel',
+ 'mistralChatModel',
])
AI_AGENT_TYPES: FrozenSet[str] = frozenset([
@@ -322,6 +325,12 @@ def detect_ai_provider(node_type: str, parameters: dict = None) -> str:
# AI Agent types get provider from parameters
if node_type in AI_AGENT_TYPES:
return (parameters or {}).get('provider', 'openai')
+ elif 'deepseek' in node_type.lower():
+ return 'deepseek'
+ elif 'kimi' in node_type.lower():
+ return 'kimi'
+ elif 'mistral' in node_type.lower():
+ return 'mistral'
elif 'cerebras' in node_type.lower():
return 'cerebras'
elif 'groq' in node_type.lower():
diff --git a/server/routers/websocket.py b/server/routers/websocket.py
index f87e1fe1..a7584d49 100644
--- a/server/routers/websocket.py
+++ b/server/routers/websocket.py
@@ -991,9 +991,9 @@ async def handle_validate_api_key(data: Dict[str, Any], websocket: WebSocket) ->
broadcaster = get_status_broadcaster()
provider, api_key = data["provider"].lower(), data["api_key"].strip()
- # Service-only providers don't have models to fetch
- AI_PROVIDERS = {'openai', 'anthropic', 'gemini', 'openrouter', 'groq', 'cerebras'}
- models = await ai_service.fetch_models(provider, api_key) if provider in AI_PROVIDERS else []
+ # Fetch models for AI providers (any provider with a models_endpoint in llm_defaults.json)
+ from services.ai import PROVIDER_CONFIGS
+ models = await ai_service.fetch_models(provider, api_key) if provider in PROVIDER_CONFIGS else []
await auth_service.store_api_key(provider=provider, api_key=api_key, models=models,
session_id=data.get("session_id", "default"))
# Broadcast with hasKey and models so frontend can update reactively
@@ -2683,7 +2683,8 @@ async def handle_get_validated_ai_providers(data: Dict[str, Any], websocket: Web
auth_service = container.auth_service()
database = container.database()
- AI_PROVIDERS = ['openai', 'anthropic', 'gemini', 'groq', 'cerebras', 'openrouter']
+ from services.ai import PROVIDER_CONFIGS
+ AI_PROVIDERS = list(PROVIDER_CONFIGS.keys())
# Load popular models from llm_defaults.json
defaults_path = Path(__file__).parent.parent / "config" / "llm_defaults.json"
diff --git a/server/services/ai.py b/server/services/ai.py
index ede6dc4a..38943f0d 100644
--- a/server/services/ai.py
+++ b/server/services/ai.py
@@ -171,6 +171,8 @@ class ProviderConfig:
default_model: str # Default model when none specified
models_endpoint: str # API endpoint to fetch models
models_header_fn: Callable[[str], dict] # Function to create headers
+ base_url: str = '' # Custom base URL for OpenAI-compatible providers
+ extra_headers: Optional[Dict[str, str]] = None # Default headers for API calls
@dataclass
@@ -219,6 +221,24 @@ def _cerebras_headers(api_key: str) -> dict:
return {'Authorization': f'Bearer {api_key}'}
+def _bearer_headers(api_key: str) -> dict:
+ """Generic Bearer auth headers for OpenAI-compatible providers."""
+ return {'Authorization': f'Bearer {api_key}'}
+
+
+# Map of provider -> (model_class, header_fn) for providers with custom LangChain classes.
+# Providers NOT in this map use ChatOpenAI + _bearer_headers (OpenAI-compatible).
+_PROVIDER_CLASS_MAP: Dict[str, tuple] = {
+ 'openai': (ChatOpenAI, _openai_headers),
+ 'anthropic': (ChatAnthropic, _anthropic_headers),
+ 'gemini': (None, _gemini_headers), # Lazy-loaded via _get_google_genai_class()
+ 'openrouter': (ChatOpenAI, _openrouter_headers),
+ 'groq': (ChatGroq, _groq_headers),
+}
+if CEREBRAS_AVAILABLE:
+ _PROVIDER_CLASS_MAP['cerebras'] = (ChatCerebras, _cerebras_headers)
+
+
# =============================================================================
# LLM DEFAULTS CONFIGURATION - Load from config/llm_defaults.json
# =============================================================================
@@ -245,9 +265,10 @@ def _get_default_model(provider: str, fallback: str) -> str:
def _resolve_max_tokens(flattened: dict, model: str, provider: str) -> int:
- """Resolve max_tokens: user param -> model registry -> llm_defaults -> 4096.
+ """Resolve max_tokens: user param (clamped to model max) -> provider default -> 4096.
- Clamps to model's actual maximum if user value exceeds it.
+ When user sets max_tokens explicitly, clamp to model's actual maximum.
+ When user doesn't set it, use the provider's default from llm_defaults.json.
"""
from services.model_registry import get_model_registry
registry = get_model_registry()
@@ -260,7 +281,10 @@ def _resolve_max_tokens(flattened: dict, model: str, provider: str) -> int:
logger.info(f"[AI] Clamping max_tokens from {user_int} to {model_max} for {provider}/{model}")
return model_max
return user_int
- return model_max
+
+ # No user value: use provider default from llm_defaults.json, not model's max capacity
+ provider_default = registry._get_default_max_output_tokens(provider, "_no_match_")
+ return min(provider_default, model_max)
def _resolve_temperature(flattened: dict, model: str, provider: str, thinking_enabled: bool) -> float:
@@ -293,75 +317,37 @@ def _resolve_temperature(flattened: dict, model: str, provider: str, thinking_en
return max(temp_range[0], min(temp_range[1], user_temp))
-# Provider configurations - defaults from config/llm_defaults.json
-PROVIDER_CONFIGS: Dict[str, ProviderConfig] = {
- 'openai': ProviderConfig(
- name='openai',
- model_class=ChatOpenAI,
- api_key_param='openai_api_key',
- max_tokens_param='max_tokens',
- detection_patterns=('gpt', 'openai', 'o1', 'o3', 'o4'),
- default_model=_get_default_model('openai', 'gpt-5.2'),
- models_endpoint='https://api.openai.com/v1/models',
- models_header_fn=_openai_headers
- ),
- 'anthropic': ProviderConfig(
- name='anthropic',
- model_class=ChatAnthropic,
- api_key_param='anthropic_api_key',
- max_tokens_param='max_tokens',
- detection_patterns=('claude', 'anthropic'),
- default_model=_get_default_model('anthropic', 'claude-opus-4-6'),
- models_endpoint='https://api.anthropic.com/v1/models',
- models_header_fn=_anthropic_headers
- ),
- 'gemini': ProviderConfig(
- name='gemini',
- model_class=None, # Lazy: resolved via _get_google_genai_class() to avoid import hang on Windows
- api_key_param='google_api_key',
- max_tokens_param='max_output_tokens',
- detection_patterns=('gemini', 'google'),
- default_model=_get_default_model('gemini', 'gemini-2.5-flash'),
- models_endpoint='https://generativelanguage.googleapis.com/v1beta/models',
- models_header_fn=_gemini_headers
- ),
- 'openrouter': ProviderConfig(
- name='openrouter',
- model_class=ChatOpenAI,
- api_key_param='api_key',
- max_tokens_param='max_tokens',
- detection_patterns=('openrouter',),
- default_model=_get_default_model('openrouter', 'anthropic/claude-opus-4.6'),
- models_endpoint='https://openrouter.ai/api/v1/models',
- models_header_fn=_openrouter_headers
- ),
- 'groq': ProviderConfig(
- name='groq',
- model_class=ChatGroq,
- api_key_param='api_key',
- max_tokens_param='max_tokens',
- detection_patterns=('groq',),
- default_model=_get_default_model('groq', 'llama-4-scout'),
- models_endpoint='https://api.groq.com/openai/v1/models',
- models_header_fn=_groq_headers
- ),
-}
+# Provider configurations - built from config/llm_defaults.json
+# Custom model_class/header_fn from _PROVIDER_CLASS_MAP; everything else from JSON.
+def _build_provider_configs() -> Dict[str, ProviderConfig]:
+ """Build PROVIDER_CONFIGS from llm_defaults.json + _PROVIDER_CLASS_MAP."""
+ providers = _LLM_DEFAULTS.get("providers", {})
+ configs: Dict[str, ProviderConfig] = {}
-# Add Cerebras only if available (requires Python <3.13)
-if CEREBRAS_AVAILABLE:
- PROVIDER_CONFIGS['cerebras'] = ProviderConfig(
- name='cerebras',
- model_class=ChatCerebras,
- api_key_param='api_key',
- max_tokens_param='max_tokens',
- detection_patterns=('cerebras', 'llama3.1', 'llama-3.3', 'qwen-3', 'qwen3', 'gpt-oss'),
- default_model=_get_default_model('cerebras', 'llama3.1-8b'),
- models_endpoint='https://api.cerebras.ai/v1/models',
- models_header_fn=_cerebras_headers
- )
- logger.info("Cerebras provider enabled")
-else:
- logger.warning(f"Cerebras provider not available: {CEREBRAS_IMPORT_ERROR or 'langchain-cerebras import failed'}")
+ for name, prov in providers.items():
+ # Skip cerebras if not available
+ if name == 'cerebras' and not CEREBRAS_AVAILABLE:
+ logger.warning(f"Cerebras provider not available: {CEREBRAS_IMPORT_ERROR or 'langchain-cerebras import failed'}")
+ continue
+
+ model_class, header_fn = _PROVIDER_CLASS_MAP.get(name, (ChatOpenAI, _bearer_headers))
+ configs[name] = ProviderConfig(
+ name=name,
+ model_class=model_class,
+ api_key_param=prov.get('api_key_param', 'api_key'),
+ max_tokens_param=prov.get('max_tokens_param', 'max_tokens'),
+ detection_patterns=tuple(prov.get('detection_patterns', [name])),
+ default_model=prov.get('default_model', ''),
+ models_endpoint=prov.get('models_endpoint', ''),
+ models_header_fn=header_fn,
+ base_url=prov.get('base_url', ''),
+ extra_headers=prov.get('extra_headers'),
+ )
+
+ return configs
+
+
+PROVIDER_CONFIGS: Dict[str, ProviderConfig] = _build_provider_configs()
def detect_provider_from_model(model: str) -> str:
@@ -375,6 +361,9 @@ def detect_provider_from_model(model: str) -> str:
def is_model_valid_for_provider(model: str, provider: str) -> bool:
"""Check if model name matches the provider's patterns."""
+ # OpenRouter is a proxy supporting all models (provider/model format) - always valid
+ if provider == 'openrouter':
+ return True
config = PROVIDER_CONFIGS.get(provider)
if not config:
return True
@@ -1181,6 +1170,14 @@ class AIService:
raise ValueError(error_msg)
raise ValueError(f"Unsupported provider: {provider}")
+ # Strip [FREE] prefix if present (added by OpenRouter model list for display)
+ if model.startswith('[FREE] '):
+ model = model[7:]
+
+ # Strip owner prefix for non-OpenRouter providers (e.g. "openai/gpt-oss-120b" → "gpt-oss-120b")
+ if provider != 'openrouter' and '/' in model:
+ model = model.split('/', 1)[-1]
+
# Build kwargs dynamically from registry config
kwargs = {
config.api_key_param: api_key,
@@ -1195,13 +1192,11 @@ class AIService:
kwargs['base_url'] = proxy_url
kwargs[config.api_key_param] = "ollama" # Ollama-style token
logger.info(f"[AI] Using proxy for {provider}: {proxy_url}")
- # OpenRouter uses OpenAI-compatible API with custom base_url
- elif provider == 'openrouter':
- kwargs['base_url'] = 'https://openrouter.ai/api/v1'
- kwargs['default_headers'] = {
- 'HTTP-Referer': 'http://localhost:3000',
- 'X-Title': 'MachinaOS'
- }
+ # OpenAI-compatible providers: apply base_url from config
+ elif config.base_url and config.model_class == ChatOpenAI and provider != 'openai':
+ kwargs['base_url'] = config.base_url
+ if config.extra_headers:
+ kwargs['default_headers'] = config.extra_headers
# Apply thinking/reasoning configuration using model registry
# The registry determines thinking_type based on model metadata
@@ -1243,6 +1238,10 @@ class AIService:
# Groq Qwen/QwQ: reasoning_format parameter
format_val = thinking.format if thinking.format in ('parsed', 'hidden') else 'parsed'
kwargs['reasoning_format'] = format_val
+ elif thinking_type == 'enabled':
+ # DeepSeek/Kimi: toggle thinking via model_kwargs extra_body
+ kwargs.setdefault('model_kwargs', {})['extra_body'] = {"thinking": {"type": "enabled"}}
+ logger.info(f"[AI] {provider} thinking: enabled via extra_body")
elif thinking_type == 'none':
logger.warning(f"[AI] Model '{model}' ({provider}) may not support thinking mode")
@@ -1279,19 +1278,33 @@ class AIService:
if not config:
raise ValueError(f"Unsupported provider: {provider}")
- curated = self._get_curated_models(provider)
endpoint = config.models_endpoint
headers = config.models_header_fn(api_key)
+ curated = self._get_curated_models(provider)
- async with httpx.AsyncClient(timeout=self.settings.ai_timeout) as client:
- url = endpoint
- response = await client.get(url, headers=headers)
- response.raise_for_status()
-
- if curated:
- return curated
+ try:
+ async with httpx.AsyncClient(timeout=self.settings.ai_timeout) as client:
+ response = await client.get(endpoint, headers=headers)
+ response.raise_for_status()
+ data = response.json()
+
+ # Both Groq and Cerebras use OpenAI-compatible /v1/models format.
+ # Their API may return owner-prefixed IDs (e.g. "openai/gpt-oss-120b")
+ # but their chat API expects flat IDs (e.g. "gpt-oss-120b"), so strip prefixes.
+ raw_ids = [m["id"] for m in data.get("data", []) if m.get("id")]
+ api_models = sorted(set(
+ mid.split("/", 1)[-1] if "/" in mid else mid
+ for mid in raw_ids
+ ))
+ if api_models:
+ return api_models
+ except Exception as e:
+ logger.warning(f"[AI] Failed to fetch models from {provider} API: {e}")
- return [config.default_model]
+ # Fallback to curated list from llm_defaults.json
+ if curated:
+ return curated
+ return [config.default_model]
async def execute_chat(self, node_id: str, node_type: str, parameters: Dict[str, Any]) -> Dict[str, Any]:
"""Execute AI chat model."""
@@ -1323,15 +1336,13 @@ class AIService:
raise ValueError("Prompt cannot be empty")
# Determine provider from node_type (more reliable than model name detection)
- # OpenRouter models have format like "openai/gpt-4o" which would incorrectly detect as openai
- if node_type == 'openrouterChatModel':
- provider = 'openrouter'
- elif node_type == 'groqChatModel':
- provider = 'groq'
- elif node_type == 'cerebrasChatModel':
- provider = 'cerebras'
- else:
- provider = self.detect_provider(model)
+ from constants import detect_ai_provider
+ provider = detect_ai_provider(node_type, flattened)
+
+ # Some APIs return owner-prefixed model IDs (e.g. "openai/gpt-oss-120b")
+ # but chat API expects flat IDs — strip prefix for non-OpenRouter providers
+ if provider != 'openrouter' and '/' in model:
+ model = model.split('/', 1)[-1]
# Build thinking config from parameters
thinking_config = None
@@ -1519,6 +1530,9 @@ class AIService:
api_key = flattened.get('api_key') or flattened.get('apiKey')
provider = parameters.get('provider', 'openai')
model = parameters.get('model', '')
+ # Strip [FREE] prefix if present (added by OpenRouter model list for display)
+ if model.startswith('[FREE] '):
+ model = model[7:]
logger.debug(f"[LangGraph] Agent: {provider}/{model}, tools={len(tool_data) if tool_data else 0}")
@@ -2014,6 +2028,9 @@ class AIService:
api_key = flattened.get('api_key') or flattened.get('apiKey')
provider = parameters.get('provider', 'openai')
model = parameters.get('model', '')
+ # Strip [FREE] prefix if present (added by OpenRouter model list for display)
+ if model.startswith('[FREE] '):
+ model = model[7:]
logger.debug(f"[ChatAgent] Provider: {provider}, Model: {model}")
diff --git a/server/services/llm/config.py b/server/services/llm/config.py
index 99f5f8ce..92044102 100644
--- a/server/services/llm/config.py
+++ b/server/services/llm/config.py
@@ -28,6 +28,7 @@ class ProviderConfig:
api_key_header: str # e.g. "Authorization", "x-api-key"
api_key_format: str = "Bearer {key}" # how the header value is built
extra_headers: Dict[str, str] = field(default_factory=dict)
+ base_url: str = "" # OpenAI-compatible base URL (e.g. "https://api.deepseek.com")
# ---------------------------------------------------------------------------
@@ -54,55 +55,43 @@ def reload_defaults() -> None:
# ---------------------------------------------------------------------------
-# Provider registry -- built from llm_defaults.json
+# Provider-specific auth overrides (most providers use Bearer auth)
# ---------------------------------------------------------------------------
-PROVIDER_CONFIGS: Dict[str, ProviderConfig] = {
- "openai": ProviderConfig(
- name="openai",
- default_model=LLM_DEFAULTS.get("providers", {}).get("openai", {}).get("default_model", "gpt-5.2"),
- detection_patterns=("gpt", "openai", "o1", "o3", "o4"),
- models_endpoint="https://api.openai.com/v1/models",
- api_key_header="Authorization",
- api_key_format="Bearer {key}",
- ),
- "anthropic": ProviderConfig(
- name="anthropic",
- default_model=LLM_DEFAULTS.get("providers", {}).get("anthropic", {}).get("default_model", "claude-sonnet-4-6"),
- detection_patterns=("claude", "anthropic"),
- models_endpoint="https://api.anthropic.com/v1/models",
- api_key_header="x-api-key",
- api_key_format="{key}",
- extra_headers={"anthropic-version": "2023-06-01"},
- ),
- "gemini": ProviderConfig(
- name="gemini",
- default_model=LLM_DEFAULTS.get("providers", {}).get("gemini", {}).get("default_model", "gemini-2.5-flash"),
- detection_patterns=("gemini", "google"),
- models_endpoint="https://generativelanguage.googleapis.com/v1beta/models",
- api_key_header="", # API key in URL query param for Gemini
- api_key_format="",
- ),
- "openrouter": ProviderConfig(
- name="openrouter",
- default_model=LLM_DEFAULTS.get("providers", {}).get("openrouter", {}).get("default_model", "anthropic/claude-sonnet-4.6"),
- detection_patterns=("openrouter",),
- models_endpoint="https://openrouter.ai/api/v1/models",
- api_key_header="Authorization",
- api_key_format="Bearer {key}",
- extra_headers={"HTTP-Referer": "http://localhost:3000", "X-Title": "MachinaOS"},
- ),
- "xai": ProviderConfig(
- name="xai",
- default_model=LLM_DEFAULTS.get("providers", {}).get("xai", {}).get("default_model", "grok-3"),
- detection_patterns=("grok", "xai"),
- models_endpoint="https://api.x.ai/v1/models",
- api_key_header="Authorization",
- api_key_format="Bearer {key}",
- ),
+_AUTH_OVERRIDES: Dict[str, Dict[str, str]] = {
+ "anthropic": {"api_key_header": "x-api-key", "api_key_format": "{key}"},
+ "gemini": {"api_key_header": "", "api_key_format": ""}, # API key in URL query param
}
+# ---------------------------------------------------------------------------
+# Provider registry -- built dynamically from llm_defaults.json
+# ---------------------------------------------------------------------------
+
+def _build_provider_configs() -> Dict[str, ProviderConfig]:
+ """Build ProviderConfig entries from llm_defaults.json."""
+ providers = LLM_DEFAULTS.get("providers", {})
+ configs: Dict[str, ProviderConfig] = {}
+
+ for name, prov in providers.items():
+ auth = _AUTH_OVERRIDES.get(name, {})
+ configs[name] = ProviderConfig(
+ name=name,
+ default_model=prov.get("default_model", ""),
+ detection_patterns=tuple(prov.get("detection_patterns", [name])),
+ models_endpoint=prov.get("models_endpoint", ""),
+ api_key_header=auth.get("api_key_header", "Authorization"),
+ api_key_format=auth.get("api_key_format", "Bearer {key}"),
+ extra_headers=prov.get("extra_headers", {}),
+ base_url=prov.get("base_url", ""),
+ )
+
+ return configs
+
+
+PROVIDER_CONFIGS: Dict[str, ProviderConfig] = _build_provider_configs()
+
+
def get_provider_config(provider: str) -> Optional[ProviderConfig]:
return PROVIDER_CONFIGS.get(provider)
diff --git a/server/services/llm/factory.py b/server/services/llm/factory.py
index 8c478183..cbf2d15c 100644
--- a/server/services/llm/factory.py
+++ b/server/services/llm/factory.py
@@ -9,8 +9,11 @@ from services.llm.protocol import LLMProvider
logger = get_logger(__name__)
-# Providers that have native SDK implementations
-NATIVE_PROVIDERS = frozenset({"anthropic", "openai", "gemini", "openrouter", "xai"})
+# Providers with dedicated SDK implementations
+_DEDICATED_PROVIDERS = frozenset({"anthropic", "openai", "gemini", "openrouter"})
+
+# All providers supported by native SDK path (dedicated + OpenAI-compatible via base_url)
+NATIVE_PROVIDERS = _DEDICATED_PROVIDERS | frozenset({"xai", "deepseek", "kimi", "mistral"})
def create_provider(
@@ -21,7 +24,8 @@ def create_provider(
) -> LLMProvider:
"""Create a native LLM provider instance.
- Lazy-imports the provider module to avoid loading SDKs at startup.
+ Dedicated providers (anthropic, gemini, openrouter) use their own classes.
+ OpenAI-compatible providers use OpenAIProvider with base_url from config.
"""
if provider == "anthropic":
from services.llm.providers.anthropic import AnthropicProvider
@@ -39,10 +43,12 @@ def create_provider(
from services.llm.providers.openrouter import OpenRouterProvider
return OpenRouterProvider(api_key, proxy_url=proxy_url)
- if provider == "xai":
- # xAI uses OpenAI-compatible API
+ # OpenAI-compatible providers: use OpenAIProvider with base_url from config
+ from services.llm.config import get_provider_config
+ config = get_provider_config(provider)
+ if config and config.base_url:
from services.llm.providers.openai import OpenAIProvider
- return OpenAIProvider(api_key, base_url="https://api.x.ai/v1", proxy_url=proxy_url)
+ return OpenAIProvider(api_key, base_url=config.base_url, proxy_url=proxy_url)
raise ValueError(f"Unknown provider: {provider}")
diff --git a/server/services/llm/providers/openai.py b/server/services/llm/providers/openai.py
index a340f6a9..84490fed 100644
--- a/server/services/llm/providers/openai.py
+++ b/server/services/llm/providers/openai.py
@@ -78,16 +78,9 @@ class OpenAIProvider:
# ------------------------------------------------------------------
async def fetch_models(self, api_key: str) -> List[str]:
- import httpx
- async with httpx.AsyncClient() as client:
- r = await client.get(
- "https://api.openai.com/v1/models",
- headers={"Authorization": f"Bearer {api_key}"},
- timeout=15.0,
- )
- r.raise_for_status()
- data = r.json()
- return sorted([m["id"] for m in data.get("data", [])])
+ """Fetch models using the client's configured base_url."""
+ models = await self._client.models.list()
+ return sorted([m.id for m in models.data])
# ------------------------------------------------------------------
# internals
diff --git a/server/services/model_registry.py b/server/services/model_registry.py
index cce185cb..2b52cefb 100644
--- a/server/services/model_registry.py
+++ b/server/services/model_registry.py
@@ -48,6 +48,8 @@ THINKING_PATTERNS: List[Tuple[str, str, str]] = [
("gemini", r"gemini-(2\.5|3)", "budget"),
("groq", r"qwen3", "format"),
("cerebras", r"qwen", "budget"),
+ ("deepseek", r"deepseek-(reasoner|chat)", "enabled"),
+ ("kimi", r"kimi-k2", "enabled"),
]
# Default temperature ranges per provider
@@ -58,6 +60,9 @@ DEFAULT_TEMP_RANGES = {
"groq": (0.0, 2.0),
"cerebras": (0.0, 1.5),
"openrouter": (0.0, 2.0),
+ "deepseek": (0.0, 2.0),
+ "kimi": (0.0, 1.0),
+ "mistral": (0.0, 1.0),
}