Skip to content

Self-improving AI Employees for everyone

fix: provider params, thinking compat, model defaults, global model label

trohitg0d85829bcc8f
diff --git a/client/src/components/ui/TopToolbar.tsx b/client/src/components/ui/TopToolbar.tsx
index acdc0e92..2b7eea0a 100644
--- a/client/src/components/ui/TopToolbar.tsx
+++ b/client/src/components/ui/TopToolbar.tsx
@@ -428,6 +428,12 @@ const TopToolbar: React.FC<TopToolbarProps> = ({
{/* Global Model Selector */}
{globalModelState.providers.length > 0 && (
<div style={{ display: 'flex', alignItems: 'center', gap: theme.spacing.sm }}>
+ <span style={{
+ fontSize: theme.fontSize.sm,
+ color: theme.dracula.cyan,
+ fontWeight: theme.fontWeight.semibold,
+ whiteSpace: 'nowrap',
+ }}>Set Global Model</span>
<Select
value={globalSelectValue}
onChange={handleSelectGlobalModel}
diff --git a/client/src/nodeDefinitions/aiAgentNodes.ts b/client/src/nodeDefinitions/aiAgentNodes.ts
index 418223e1..b1f4b891 100644
--- a/client/src/nodeDefinitions/aiAgentNodes.ts
+++ b/client/src/nodeDefinitions/aiAgentNodes.ts
@@ -3,6 +3,7 @@ import {
INodeTypeDescription,
NodeConnectionType
} from '../types/INodeProperties';
+import { AI_PROVIDER_OPTIONS } from './aiModelNodes';
// ============================================================================
// AI AGENT AND CHAT NODES
@@ -62,24 +63,7 @@ export const aiAgentNodes: Record<string, INodeTypeDescription> = {
displayName: 'AI Provider',
name: 'provider',
type: 'options',
- options: [
- {
- name: 'OpenAI',
- value: 'openai'
- },
- {
- name: 'Anthropic (Claude)',
- value: 'anthropic'
- },
- {
- name: 'Google (Gemini)',
- value: 'gemini'
- },
- {
- name: 'OpenRouter',
- value: 'openrouter'
- }
- ],
+ options: AI_PROVIDER_OPTIONS,
default: 'openai',
description: 'The AI provider to use (configure API keys in Credentials)'
},
@@ -228,24 +212,7 @@ export const aiAgentNodes: Record<string, INodeTypeDescription> = {
displayName: 'AI Provider',
name: 'provider',
type: 'options',
- options: [
- {
- name: 'OpenAI',
- value: 'openai'
- },
- {
- name: 'Anthropic (Claude)',
- value: 'anthropic'
- },
- {
- name: 'Google (Gemini)',
- value: 'gemini'
- },
- {
- name: 'OpenRouter',
- value: 'openrouter'
- }
- ],
+ options: AI_PROVIDER_OPTIONS,
default: 'openai',
description: 'The AI provider to use (configure API keys in Credentials)'
},
diff --git a/client/src/nodeDefinitions/aiModelNodes.ts b/client/src/nodeDefinitions/aiModelNodes.ts
index f31fcf94..1228fcbe 100644
--- a/client/src/nodeDefinitions/aiModelNodes.ts
+++ b/client/src/nodeDefinitions/aiModelNodes.ts
@@ -187,7 +187,7 @@ const deepseekConfig: ChatModelConfig = {
displayName: 'DeepSeek',
icon: DEEPSEEK_ICON,
color: dracula.cyan,
- description: 'DeepSeek V3 models with Chain-of-Thought reasoning (deepseek-chat, deepseek-reasoner)',
+ description: 'DeepSeek V3 models (deepseek-chat, deepseek-reasoner with always-on CoT)',
models: [],
parameters: [
{
@@ -200,7 +200,6 @@ const deepseekConfig: ChatModelConfig = {
STANDARD_PARAMETERS.presencePenalty,
STANDARD_PARAMETERS.timeout,
STANDARD_PARAMETERS.maxRetries,
- STANDARD_PARAMETERS.thinkingEnabled,
]
};
@@ -210,7 +209,7 @@ const kimiConfig: ChatModelConfig = {
displayName: 'Kimi',
icon: KIMI_ICON,
color: dracula.purple,
- description: 'Kimi K2 models by Moonshot AI with 256K context and reasoning support',
+ description: 'Kimi K2 models by Moonshot AI with 256K context (thinking on by default)',
models: [],
parameters: [
{
@@ -222,7 +221,6 @@ const kimiConfig: ChatModelConfig = {
STANDARD_PARAMETERS.topP,
STANDARD_PARAMETERS.timeout,
STANDARD_PARAMETERS.maxRetries,
- STANDARD_PARAMETERS.thinkingEnabled,
]
};
@@ -272,5 +270,12 @@ export const AI_MODEL_PROVIDER_MAP: Record<string, string> = Object.fromEntries(
ALL_CONFIGS.map(c => [`${c.providerId}ChatModel`, c.providerId])
);
+// AI provider options for agent node dropdowns -- derived from AI_PROVIDER_META (single source of truth)
+import { AI_PROVIDER_META } from '../components/icons/AIProviderIcons';
+export const AI_PROVIDER_OPTIONS = Object.entries(AI_PROVIDER_META).map(([id, meta]) => ({
+ name: meta.label,
+ value: id,
+}));
+
// Export configurations and factory for external use
export { openaiConfig, claudeConfig, geminiConfig, openrouterConfig, groqConfig, cerebrasConfig, deepseekConfig, kimiConfig, mistralConfig, createBaseChatModel };
diff --git a/client/src/nodeDefinitions/specializedAgentNodes.ts b/client/src/nodeDefinitions/specializedAgentNodes.ts
index 77777d9f..20583944 100644
--- a/client/src/nodeDefinitions/specializedAgentNodes.ts
+++ b/client/src/nodeDefinitions/specializedAgentNodes.ts
@@ -5,6 +5,7 @@ import {
INodeProperties
} from '../types/INodeProperties';
import { dracula } from '../styles/theme';
+import { AI_PROVIDER_OPTIONS } from './aiModelNodes';
// ============================================================================
// SHARED AI AGENT INPUTS - Used by Specialized Agent Nodes
@@ -64,12 +65,7 @@ export const AI_AGENT_PROPERTIES: INodeProperties[] = [
displayName: 'AI Provider',
name: 'provider',
type: 'options',
- options: [
- { name: 'OpenAI', value: 'openai' },
- { name: 'Anthropic (Claude)', value: 'anthropic' },
- { name: 'Google (Gemini)', value: 'gemini' },
- { name: 'OpenRouter', value: 'openrouter' }
- ],
+ options: AI_PROVIDER_OPTIONS,
default: 'openai',
description: 'The AI provider to use (configure API keys in Credentials)'
},
diff --git a/server/config/llm_defaults.json b/server/config/llm_defaults.json
index e8aac9e5..ebc41516 100644
--- a/server/config/llm_defaults.json
+++ b/server/config/llm_defaults.json
@@ -173,6 +173,7 @@
"models_endpoint": "https://api.deepseek.com/models",
"api_key_param": "api_key",
"max_tokens_param": "max_tokens",
+ "supported_params": ["temperature", "max_tokens", "top_p", "frequency_penalty", "presence_penalty", "stop", "seed", "tools", "tool_choice", "response_format"],
"max_output_tokens": {
"deepseek-chat": 8192,
"deepseek-reasoner": 65536,
@@ -183,8 +184,7 @@
"deepseek-reasoner": 131072,
"_default": 131072
},
- "thinking_models": ["deepseek-reasoner", "deepseek-chat"],
- "thinking_type": "enabled",
+ "reasoning_models": ["deepseek-reasoner"],
"temperature_range": [0.0, 2.0]
},
"kimi": {
@@ -194,24 +194,19 @@
"models_endpoint": "https://api.moonshot.ai/v1/models",
"api_key_param": "api_key",
"max_tokens_param": "max_tokens",
+ "supported_params": ["temperature", "max_tokens", "top_p", "frequency_penalty", "presence_penalty", "stop", "seed", "tools", "tool_choice", "response_format"],
+ "thinking_default_on": ["kimi-k2.5", "kimi-k2-thinking", "kimi-k2-0905-preview"],
+ "fixed_temperature": {"kimi-k2.5": 0.6, "kimi-k2-thinking": 1.0, "kimi-k2-0905-preview": 0.6},
"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": {
@@ -221,21 +216,20 @@
"models_endpoint": "https://api.mistral.ai/v1/models",
"api_key_param": "api_key",
"max_tokens_param": "max_tokens",
+ "supported_params": ["temperature", "max_tokens", "top_p", "frequency_penalty", "presence_penalty", "stop", "seed", "tools", "tool_choice", "response_format"],
"max_output_tokens": {
"mistral-large-latest": 131072,
"mistral-small-latest": 32768,
- "codestral-latest": 32768,
- "open-mistral-nemo": 131072,
+ "codestral-latest": 131072,
"_default": 8192
},
"context_length": {
"mistral-large-latest": 262144,
"mistral-small-latest": 32768,
- "codestral-latest": 262144,
- "open-mistral-nemo": 131072,
+ "codestral-latest": 256000,
"_default": 131072
},
- "temperature_range": [0.0, 1.0]
+ "temperature_range": [0.0, 1.5]
}
}
}
diff --git a/server/config/model_registry.json b/server/config/model_registry.json
index a70ba74f..ccb1456a 100644
--- a/server/config/model_registry.json
+++ b/server/config/model_registry.json
@@ -1,8 +1,71 @@
{
- "_generated": "2026-04-02T10:34:48.408712+00:00",
+ "_generated": "2026-04-03T10:36:19.816708+00:00",
"_source": "openrouter",
- "_model_count": 350,
+ "_model_count": 349,
"models": {
+ "gemini/gemma-4-31b-it": {
+ "id": "google/gemma-4-31b-it",
+ "name": "Google: Gemma 4 31B",
+ "provider": "gemini",
+ "local_id": "gemma-4-31b-it",
+ "context_length": 262144,
+ "max_output_tokens": 131072,
+ "input_price_per_mtok": 0.14,
+ "output_price_per_mtok": 0.4,
+ "supports_thinking": false,
+ "thinking_type": "none",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "frequency_penalty",
+ "include_reasoning",
+ "logit_bias",
+ "max_tokens",
+ "presence_penalty",
+ "reasoning",
+ "repetition_penalty",
+ "response_format",
+ "seed",
+ "stop",
+ "structured_outputs",
+ "temperature",
+ "top_k",
+ "top_p"
+ ]
+ },
+ "qwen/qwen3.6-plus:free": {
+ "id": "qwen/qwen3.6-plus:free",
+ "name": "Qwen: Qwen3.6 Plus (free)",
+ "provider": "qwen",
+ "local_id": "qwen3.6-plus:free",
+ "context_length": 1000000,
+ "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": [
+ "include_reasoning",
+ "max_tokens",
+ "presence_penalty",
+ "reasoning",
+ "response_format",
+ "seed",
+ "structured_outputs",
+ "temperature",
+ "tool_choice",
+ "tools",
+ "top_p"
+ ]
+ },
"z-ai/glm-5v-turbo": {
"id": "z-ai/glm-5v-turbo",
"name": "Z.ai: GLM 5V Turbo",
@@ -36,9 +99,9 @@
"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,
+ "max_output_tokens": 262144,
+ "input_price_per_mtok": 0.22,
+ "output_price_per_mtok": 0.85,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -47,9 +110,17 @@
],
"is_reasoning_model": false,
"supported_parameters": [
+ "frequency_penalty",
"include_reasoning",
+ "logit_bias",
"max_tokens",
+ "presence_penalty",
"reasoning",
+ "repetition_penalty",
+ "response_format",
+ "seed",
+ "stop",
+ "structured_outputs",
"temperature",
"tool_choice",
"tools",
@@ -165,36 +236,6 @@
"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",
@@ -1205,10 +1246,10 @@
"name": "MiniMax: MiniMax M2.5",
"provider": "minimax",
"local_id": "minimax-m2.5",
- "context_length": 196600,
- "max_output_tokens": 196600,
+ "context_length": 196608,
+ "max_output_tokens": 65536,
"input_price_per_mtok": 0.118,
- "output_price_per_mtok": 1.25,
+ "output_price_per_mtok": 0.99,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -1503,7 +1544,7 @@
"context_length": 262144,
"max_output_tokens": 262144,
"input_price_per_mtok": 0.3827,
- "output_price_per_mtok": 1.9092,
+ "output_price_per_mtok": 1.72,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -3269,64 +3310,6 @@
"top_p"
]
},
- "liquid/lfm2-8b-a1b": {
- "id": "liquid/lfm2-8b-a1b",
- "name": "LiquidAI: LFM2-8B-A1B",
- "provider": "liquid",
- "local_id": "lfm2-8b-a1b",
- "context_length": 32768,
- "max_output_tokens": 0,
- "input_price_per_mtok": 0.01,
- "output_price_per_mtok": 0.02,
- "supports_thinking": false,
- "thinking_type": "none",
- "temperature_range": [
- 0.0,
- 2.0
- ],
- "is_reasoning_model": false,
- "supported_parameters": [
- "frequency_penalty",
- "max_tokens",
- "min_p",
- "presence_penalty",
- "repetition_penalty",
- "seed",
- "stop",
- "temperature",
- "top_k",
- "top_p"
- ]
- },
- "liquid/lfm-2.2-6b": {
- "id": "liquid/lfm-2.2-6b",
- "name": "LiquidAI: LFM2-2.6B",
- "provider": "liquid",
- "local_id": "lfm-2.2-6b",
- "context_length": 32768,
- "max_output_tokens": 0,
- "input_price_per_mtok": 0.01,
- "output_price_per_mtok": 0.02,
- "supports_thinking": false,
- "thinking_type": "none",
- "temperature_range": [
- 0.0,
- 2.0
- ],
- "is_reasoning_model": false,
- "supported_parameters": [
- "frequency_penalty",
- "max_tokens",
- "min_p",
- "presence_penalty",
- "repetition_penalty",
- "seed",
- "stop",
- "temperature",
- "top_k",
- "top_p"
- ]
- },
"ibm-granite/granite-4.0-h-micro": {
"id": "ibm-granite/granite-4.0-h-micro",
"name": "IBM: Granite 4.0 Micro",
@@ -4662,8 +4645,8 @@
"max_output_tokens": 7168,
"input_price_per_mtok": 0.15,
"output_price_per_mtok": 0.75,
- "supports_thinking": false,
- "thinking_type": "none",
+ "supports_thinking": true,
+ "thinking_type": "enabled",
"temperature_range": [
0.0,
2.0
@@ -5027,6 +5010,7 @@
"frequency_penalty",
"include_reasoning",
"logit_bias",
+ "logprobs",
"max_tokens",
"min_p",
"presence_penalty",
@@ -5041,6 +5025,7 @@
"tool_choice",
"tools",
"top_k",
+ "top_logprobs",
"top_p"
]
},
@@ -7188,8 +7173,8 @@
"max_output_tokens": 0,
"input_price_per_mtok": 0.2,
"output_price_per_mtok": 0.77,
- "supports_thinking": false,
- "thinking_type": "none",
+ "supports_thinking": true,
+ "thinking_type": "enabled",
"temperature_range": [
0.0,
2.0
@@ -8401,8 +8386,8 @@
"max_output_tokens": 163840,
"input_price_per_mtok": 0.32,
"output_price_per_mtok": 0.89,
- "supports_thinking": false,
- "thinking_type": "none",
+ "supports_thinking": true,
+ "thinking_type": "enabled",
"temperature_range": [
0.0,
2.0
diff --git a/server/services/ai.py b/server/services/ai.py
index 38943f0d..fcc23144 100644
--- a/server/services/ai.py
+++ b/server/services/ai.py
@@ -173,6 +173,7 @@ class ProviderConfig:
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
+ supported_params: Optional[List[str]] = None # API-supported params (from llm_defaults.json)
@dataclass
@@ -342,6 +343,7 @@ def _build_provider_configs() -> Dict[str, ProviderConfig]:
models_header_fn=header_fn,
base_url=prov.get('base_url', ''),
extra_headers=prov.get('extra_headers'),
+ supported_params=prov.get('supported_params'),
)
return configs
@@ -1192,11 +1194,39 @@ 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}")
- # OpenAI-compatible providers: apply base_url from config
+ # OpenAI-compatible providers: apply base_url + pass only supported params
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
+ # LangChain ChatOpenAI converts max_tokens -> max_completion_tokens (OpenAI-specific).
+ # Non-OpenAI providers reject this, so pass supported params via extra_body.
+ del kwargs[config.max_tokens_param]
+ extra_body: Dict[str, Any] = {'max_tokens': max_tokens}
+ if config.supported_params:
+ if 'temperature' not in config.supported_params:
+ kwargs.pop('temperature', None)
+ if 'frequency_penalty' not in config.supported_params:
+ kwargs.pop('frequency_penalty', None)
+ if 'presence_penalty' not in config.supported_params:
+ kwargs.pop('presence_penalty', None)
+ # Handle provider-specific constraints from llm_defaults.json
+ prov_json = _LLM_DEFAULTS.get("providers", {}).get(provider, {})
+
+ # Disable thinking for models that have it ON by default (e.g. kimi-k2.5)
+ default_on = prov_json.get("thinking_default_on", [])
+ if default_on and any(model.startswith(m) for m in default_on):
+ if not (thinking and thinking.enabled):
+ extra_body["thinking"] = {"type": "disabled"}
+
+ # Apply fixed temperature per model (e.g. kimi-k2.5 requires exactly 0.6)
+ fixed_temps = prov_json.get("fixed_temperature", {})
+ for prefix, fixed_temp in fixed_temps.items():
+ if model.startswith(prefix):
+ kwargs["temperature"] = fixed_temp
+ break
+
+ kwargs['extra_body'] = extra_body
# Apply thinking/reasoning configuration using model registry
# The registry determines thinking_type based on model metadata
@@ -1238,10 +1268,6 @@ 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")
diff --git a/server/services/llm/config.py b/server/services/llm/config.py
index 92044102..0cf020f6 100644
--- a/server/services/llm/config.py
+++ b/server/services/llm/config.py
@@ -170,6 +170,13 @@ def resolve_temperature(params: dict, model: str, provider: str, thinking_enable
if thinking_enabled and provider == "anthropic":
return 1.0
+ # Fixed temperature per model from llm_defaults.json (e.g. kimi-k2.5 = 0.6)
+ prov_json = LLM_DEFAULTS.get("providers", {}).get(provider, {})
+ fixed_temps = prov_json.get("fixed_temperature", {})
+ for prefix, fixed_temp in fixed_temps.items():
+ if model.startswith(prefix):
+ return float(fixed_temp)
+
lo, hi = registry.get_temperature_range(model, provider)
return max(lo, min(hi, user_temp))
diff --git a/server/services/model_registry.py b/server/services/model_registry.py
index 2b52cefb..38304c71 100644
--- a/server/services/model_registry.py
+++ b/server/services/model_registry.py
@@ -48,8 +48,6 @@ 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