Skip to content

Self-improving AI Employees for everyone

chore(llm): update provider model defaults + docs to current lineup (2026.06)

trohitg69b26c7bde2a
diff --git a/README.md b/README.md
index b78bb326..9ad86751 100644
--- a/README.md
+++ b/README.md
@@ -104,13 +104,13 @@ RAG pipeline out of the box: parse PDFs and HTML, chunk into searchable pieces,
| Provider | Notes |
|--------------|--------------------------------------------------------------------------|
| OpenAI | GPT-5 family, GPT-4.1, o-series reasoning models, GPT-4o |
-| Anthropic | Claude Opus 4.x, Sonnet 4.x, Haiku 4.5 — with extended thinking |
+| Anthropic | Claude Fable 5, Opus 4.x, Sonnet 4.6, Haiku 4.5 — with extended thinking |
| Google | Gemini 3 Pro/Flash, 2.5 Pro/Flash — with reasoning budgets |
-| DeepSeek | DeepSeek V3, DeepSeek Reasoner |
-| Kimi | Kimi K2.5, Kimi K2 Thinking |
-| Mistral | Mistral Large/Small, Codestral |
-| Groq | Llama 3/4, Qwen3, GPT-OSS (ultra-fast inference) |
-| Cerebras | Llama 3.1, Qwen-3-235b (custom AI hardware) |
+| DeepSeek | DeepSeek V4 (Flash/Pro); chat/reasoner legacy aliases |
+| Kimi | Kimi K2.6, K2.5, K2.7-Code |
+| Mistral | Mistral Large/Medium/Small, Codestral |
+| Groq | Llama 3.x, Qwen3, GPT-OSS (ultra-fast inference) |
+| Cerebras | GPT-OSS-120b, Qwen-3-235b, GLM-4.7 (custom AI hardware) |
| OpenRouter | 200+ models via one unified API |
| **Ollama** | Run any local model on your machine — free, private, offline |
| **LM Studio**| Run any local model with a desktop app — free, private, offline |
diff --git a/docs-internal/memory_compaction.md b/docs-internal/memory_compaction.md
index 11bb0405..391d2f56 100644
--- a/docs-internal/memory_compaction.md
+++ b/docs-internal/memory_compaction.md
@@ -457,7 +457,7 @@ if memory_data and memory_data.get('session_id'):
3. **Per-Session State**: Each memory session has independent token tracking and thresholds. This allows different agents to have different compaction settings.
-4. **Model-Aware Threshold**: Threshold is `compaction_ratio` x the model's context window (default ratio **0.8** = 80%). For example, Claude Opus 4.6 (1M context) gets an 800K threshold, GPT-5.2 (400K) gets 320K, Groq models (131K) get ~105K. Ratio precedence: per-session `custom_threshold` > per-user `UserSettings.compaction_ratio` > env `Settings.compaction_ratio` (`COMPACTION_RATIO`, default 0.8) > JSON `llm_defaults.json:agent.compaction.ratio` fallback.
+4. **Model-Aware Threshold**: Threshold is `compaction_ratio` x the model's context window (default ratio **0.8** = 80%). For example, Claude Opus 4.8 (1M context) gets an 800K threshold, GPT-5.2 (400K) gets 320K, Groq models (131K) get ~105K. Ratio precedence: per-session `custom_threshold` > per-user `UserSettings.compaction_ratio` > env `Settings.compaction_ratio` (`COMPACTION_RATIO`, default 0.8) > JSON `llm_defaults.json:agent.compaction.ratio` fallback.
5. **Singleton Pattern**: Service accessible via `get_compaction_service()` for easy integration anywhere in the codebase.
diff --git a/docs-internal/native_llm_sdk.md b/docs-internal/native_llm_sdk.md
index dfd3badd..d503676b 100644
--- a/docs-internal/native_llm_sdk.md
+++ b/docs-internal/native_llm_sdk.md
@@ -170,7 +170,7 @@ Adding a new OpenAI-compatible provider is a pure config change:
```json
"deepseek": {
- "default_model": "deepseek-chat",
+ "default_model": "deepseek-v4-flash",
"detection_patterns": ["deepseek"],
"models_endpoint": "https://api.deepseek.com/v1/models",
"base_url": "https://api.deepseek.com/v1",
diff --git a/docs-internal/node-logic-flows/ai_chat_models/anthropicChatModel.md b/docs-internal/node-logic-flows/ai_chat_models/anthropicChatModel.md
index 0f1c308e..0382ce65 100644
--- a/docs-internal/node-logic-flows/ai_chat_models/anthropicChatModel.md
+++ b/docs-internal/node-logic-flows/ai_chat_models/anthropicChatModel.md
@@ -103,7 +103,7 @@ flowchart TD
- **Model ID format**: Anthropic API uses hyphens (`claude-sonnet-4-6`), NOT dots. Dotted IDs (as used by OpenRouter or Gemini) will 404. The frontend model list already uses the correct form, but template-substituted or hand-entered IDs with dots will fail.
- **`maxTokens > thinkingBudget` required**: if thinking is enabled and `maxTokens <= thinkingBudget`, the Anthropic API rejects the request. The handler does not pre-validate this; the resulting error becomes the envelope's `error` field.
- **Temperature forced to 1 when thinking enabled**: regardless of user input.
-- **1M context beta**: Opus 4.6 / Sonnet 4.x support 1M context tokens via a beta header; configured in `llm_defaults.json` rather than the node.
+- **1M context**: current Claude models (Fable 5, Opus 4.x, Sonnet 4.6) support a 1M context window; configured in `llm_defaults.json` rather than the node.
- **Errors swallowed into envelope**: handler never raises; all failures surface as `success=false`.
- **No streaming** / **no pricing tracking** on standalone chat nodes.
diff --git a/docs-internal/node-logic-flows/ai_chat_models/deepseekChatModel.md b/docs-internal/node-logic-flows/ai_chat_models/deepseekChatModel.md
index a8c25241..f381b595 100644
--- a/docs-internal/node-logic-flows/ai_chat_models/deepseekChatModel.md
+++ b/docs-internal/node-logic-flows/ai_chat_models/deepseekChatModel.md
@@ -11,7 +11,7 @@
## Purpose
-DeepSeek V3 models (`deepseek-chat`, `deepseek-reasoner`). Uses the OpenAI-compatible DeepSeek endpoint via the `services/llm/providers` layer (native path). Shares `handle_ai_chat_model`.
+DeepSeek V4 models (`deepseek-v4-flash`, `deepseek-v4-pro`); `deepseek-chat` / `deepseek-reasoner` remain as legacy aliases (deprecate 2026-07-24). Uses the OpenAI-compatible DeepSeek endpoint via the `services/llm/providers` layer (native path). Shares `handle_ai_chat_model`.
## Inputs (handles)
@@ -25,7 +25,7 @@ DeepSeek V3 models (`deepseek-chat`, `deepseek-reasoner`). Uses the OpenAI-compa
|------|------|---------|----------|---------------------|-------------|
| `prompt` | string | `""` | yes | - | User message |
| `systemMessage` | string | `""` | no | - | System prompt |
-| `model` | string | injected | no | - | `deepseek-chat` (V3) or `deepseek-reasoner` (always-on CoT) |
+| `model` | string | injected | no | - | `deepseek-v4-flash` / `deepseek-v4-pro`; `deepseek-chat` / `deepseek-reasoner` legacy aliases (reasoner = always-on CoT) |
| `temperature` | number | 0-2 | no | - | |
| `maxTokens` | number | 8-64K | no | - | |
| `topP` | number | - | no | - | |
diff --git a/docs-internal/node-logic-flows/ai_chat_models/kimiChatModel.md b/docs-internal/node-logic-flows/ai_chat_models/kimiChatModel.md
index d46ecbde..7c6038dd 100644
--- a/docs-internal/node-logic-flows/ai_chat_models/kimiChatModel.md
+++ b/docs-internal/node-logic-flows/ai_chat_models/kimiChatModel.md
@@ -11,7 +11,7 @@
## Purpose
-Kimi K2 models by Moonshot AI (`kimi-k2.5`, `kimi-k2-thinking`). 256K context, 96K output. Uses OpenAI-compatible Moonshot endpoint via the native path. Shares `handle_ai_chat_model`.
+Kimi models by Moonshot AI (`kimi-k2.6`, `kimi-k2.5`, `kimi-k2.7-code`). 256K context, 96K output. Uses OpenAI-compatible Moonshot endpoint via the native path. Shares `handle_ai_chat_model`.
## Inputs (handles)
@@ -25,7 +25,7 @@ Kimi K2 models by Moonshot AI (`kimi-k2.5`, `kimi-k2-thinking`). 256K context, 9
|------|------|---------|----------|---------------------|-------------|
| `prompt` | string | `""` | yes | - | User message |
| `systemMessage` | string | `""` | no | - | System prompt |
-| `model` | string | injected | no | - | `kimi-k2.5` (instant) or `kimi-k2-thinking` |
+| `model` | string | injected | no | - | `kimi-k2.6`, `kimi-k2.5`, or `kimi-k2.7-code` |
| `temperature` | number | **fixed** | no | - | Fixed at 0.6 (instant) or 1.0 (thinking). Input is ignored. |
| `maxTokens` | number | up to 96K | no | - | |
| `thinkingEnabled` | boolean | true for k2.5 | no | - | ON by default for k2.5; explicitly disabled for tool-calling agents |
diff --git a/docs-internal/node-logic-flows/ai_chat_models/mistralChatModel.md b/docs-internal/node-logic-flows/ai_chat_models/mistralChatModel.md
index 504feb62..7db34193 100644
--- a/docs-internal/node-logic-flows/ai_chat_models/mistralChatModel.md
+++ b/docs-internal/node-logic-flows/ai_chat_models/mistralChatModel.md
@@ -11,7 +11,7 @@
## Purpose
-Mistral AI models (`mistral-large-latest`, `mistral-small-latest`, `codestral-latest`). Up to 256K context, 131K output. No thinking/reasoning support. Uses OpenAI-compatible Mistral endpoint via native path. Shares `handle_ai_chat_model`.
+Mistral AI models (`mistral-large-latest`, `mistral-medium-latest`, `mistral-small-latest`, `codestral-latest`). Up to 256K context, 131K output. No thinking/reasoning support. Uses OpenAI-compatible Mistral endpoint via native path. Shares `handle_ai_chat_model`.
## Inputs (handles)
@@ -25,7 +25,7 @@ Mistral AI models (`mistral-large-latest`, `mistral-small-latest`, `codestral-la
|------|------|---------|----------|---------------------|-------------|
| `prompt` | string | `""` | yes | - | User message |
| `systemMessage` | string | `""` | no | - | System prompt |
-| `model` | string | injected | no | - | `mistral-large-latest`, `mistral-small-latest`, `codestral-latest` |
+| `model` | string | injected | no | - | `mistral-large-latest`, `mistral-medium-latest`, `mistral-small-latest`, `codestral-latest` |
| `temperature` | number | 0-1.5 | no | - | Narrower range than OpenAI (0-1.5) |
| `maxTokens` | number | up to 131K | no | - | |
| `topP` | number | - | no | - | |
diff --git a/docs-internal/workflow-schema.md b/docs-internal/workflow-schema.md
index a1e1a31d..2c28e8b8 100644
--- a/docs-internal/workflow-schema.md
+++ b/docs-internal/workflow-schema.md
@@ -85,8 +85,8 @@ A workflow JSON document contains:
- `openrouterChatModel` - OpenRouter unified API (200+ models)
- `groqChatModel` - Groq ultra-fast inference (Llama, Qwen3, GPT-OSS)
- `cerebrasChatModel` - Cerebras custom AI hardware (Llama, Qwen)
-- `deepseekChatModel` - DeepSeek V3 (chat + reasoner with Chain-of-Thought)
-- `kimiChatModel` - Moonshot Kimi K2.5 / K2-thinking
+- `deepseekChatModel` - DeepSeek V4 (deepseek-v4-flash / deepseek-v4-pro; chat/reasoner legacy aliases)
+- `kimiChatModel` - Moonshot Kimi K2.6 / K2.5 / K2.7-code
- `mistralChatModel` - Mistral Large / Small / Codestral
### AI Agents and Memory (3 nodes)
diff --git a/server/config/llm_defaults.json b/server/config/llm_defaults.json
index 91b4bd10..ce07c0d1 100644
--- a/server/config/llm_defaults.json
+++ b/server/config/llm_defaults.json
@@ -1,6 +1,6 @@
{
- "version": "2026.05",
- "last_updated": "2026-05-02",
+ "version": "2026.06",
+ "last_updated": "2026-06-16",
"_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).",
"agent": {
@@ -23,47 +23,47 @@
"max_output_tokens": {
"gpt-5.5": 128000,
"gpt-5.4": 128000,
- "gpt-5.3-chat": 16384,
"gpt-5.2": 128000,
"o4-mini": 100000,
+ "o3": 100000,
"gpt-4.1": 32768,
"_default": 16384
},
"context_length": {
"gpt-5.5": 1050000,
"gpt-5.4": 1050000,
- "gpt-5.3-chat": 128000,
"gpt-5.2": 400000,
"o4-mini": 200000,
+ "o3": 200000,
"gpt-4.1": 1047576,
"_default": 128000
},
- "reasoning_models": ["o4-mini"],
+ "reasoning_models": ["o4-mini", "o3"],
"thinking_models": ["gpt-5"],
"thinking_type": "effort",
"temperature_range": [0.0, 2.0]
},
"anthropic": {
- "default_model": "claude-sonnet-4-6",
+ "default_model": "claude-opus-4-8",
"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,
+ "claude-fable-5": 128000,
+ "claude-opus-4-8": 128000,
+ "claude-opus-4-7": 128000,
+ "claude-sonnet-4-6": 64000,
"claude-haiku-4-5": 64000,
- "claude-sonnet-4-0": 64000,
- "claude-opus-4-0": 32000,
"_default": 8192
},
"context_length": {
- "claude-opus-4-6": 200000,
- "claude-sonnet-4-6": 200000,
+ "claude-fable-5": 1048576,
+ "claude-opus-4-8": 1048576,
+ "claude-opus-4-7": 1048576,
+ "claude-sonnet-4-6": 1048576,
"claude-haiku-4-5": 200000,
- "claude-sonnet-4-0": 200000,
- "claude-opus-4-0": 200000,
"_default": 200000
},
"thinking_type": "budget",
@@ -120,23 +120,23 @@
"temperature_range": [0.0, 2.0]
},
"groq": {
- "default_model": "meta-llama/llama-4-scout-17b-16e-instruct",
+ "default_model": "llama-3.3-70b-versatile",
"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": {
- "meta-llama/llama-4-scout-17b-16e-instruct": 8192,
- "meta-llama/llama-4-maverick-17b-128e-instruct": 16384,
+ "openai/gpt-oss-120b": 65536,
+ "openai/gpt-oss-20b": 65536,
"llama-3.3-70b-versatile": 32768,
"llama-3.1-8b-instant": 131072,
"qwen/qwen3-32b": 40960,
"_default": 8192
},
"context_length": {
- "meta-llama/llama-4-scout-17b-16e-instruct": 131072,
- "meta-llama/llama-4-maverick-17b-128e-instruct": 131072,
+ "openai/gpt-oss-120b": 131072,
+ "openai/gpt-oss-20b": 131072,
"llama-3.3-70b-versatile": 131072,
"llama-3.1-8b-instant": 131072,
"qwen/qwen3-32b": 131072,
@@ -148,48 +148,50 @@
},
"cerebras": {
"default_model": "gpt-oss-120b",
- "detection_patterns": ["cerebras", "gpt-oss", "llama3.1", "qwen-3-235b"],
+ "detection_patterns": ["cerebras", "gpt-oss", "qwen-3-235b", "zai-glm"],
"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": {
- "gpt-oss-120b": 8192,
- "qwen-3-235b": 8192,
- "llama3.1-8b": 8192,
+ "gpt-oss-120b": 65536,
+ "qwen-3-235b": 40960,
+ "zai-glm-4.7": 40960,
"_default": 8192
},
"context_length": {
- "gpt-oss-120b": 128000,
- "qwen-3-235b": 32768,
- "llama3.1-8b": 128000,
- "_default": 128000
+ "gpt-oss-120b": 131072,
+ "qwen-3-235b": 131072,
+ "zai-glm-4.7": 131072,
+ "_default": 131072
},
"thinking_models": ["qwen-3-235b"],
"thinking_type": "budget",
"temperature_range": [0.0, 1.5]
},
"xai": {
- "default_model": "grok-4.3",
+ "default_model": "grok-4.20",
"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-4.20": 131072,
"grok-4.3": 131072,
"grok-3": 131072,
"_default": 8192
},
"context_length": {
- "grok-4.3": 256000,
+ "grok-4.20": 2000000,
+ "grok-4.3": 1000000,
"grok-3": 131072,
"_default": 131072
},
"temperature_range": [0.0, 2.0]
},
"deepseek": {
- "default_model": "deepseek-chat",
+ "default_model": "deepseek-v4-flash",
"detection_patterns": ["deepseek"],
"base_url": "https://api.deepseek.com",
"models_endpoint": "https://api.deepseek.com/models",
@@ -197,11 +199,15 @@
"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-v4-flash": 65536,
+ "deepseek-v4-pro": 65536,
"deepseek-chat": 8192,
"deepseek-reasoner": 65536,
"_default": 8192
},
"context_length": {
+ "deepseek-v4-flash": 1048576,
+ "deepseek-v4-pro": 1048576,
"deepseek-chat": 131072,
"deepseek-reasoner": 131072,
"_default": 131072
@@ -217,18 +223,18 @@
"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-k2.6", "kimi-k2.5", "kimi-k2-thinking", "kimi-k2-0905-preview"],
- "fixed_temperature": {"kimi-k2.6": 0.6, "kimi-k2.5": 0.6, "kimi-k2-thinking": 1.0, "kimi-k2-0905-preview": 0.6},
+ "thinking_default_on": ["kimi-k2.6", "kimi-k2.5", "kimi-k2.7-code"],
+ "fixed_temperature": {"kimi-k2.6": 0.6, "kimi-k2.5": 0.6, "kimi-k2.7-code": 0.6},
"max_output_tokens": {
"kimi-k2.6": 32768,
"kimi-k2.5": 98304,
- "kimi-k2-thinking": 98304,
+ "kimi-k2.7-code": 98304,
"_default": 8192
},
"context_length": {
"kimi-k2.6": 262144,
"kimi-k2.5": 262144,
- "kimi-k2-thinking": 262144,
+ "kimi-k2.7-code": 262144,
"_default": 262144
},
"temperature_range": [0.0, 1.0]
@@ -243,12 +249,14 @@
"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-medium-latest": 131072,
"mistral-small-latest": 32768,
"codestral-latest": 131072,
"_default": 8192
},
"context_length": {
"mistral-large-latest": 262144,
+ "mistral-medium-latest": 262144,
"mistral-small-latest": 32768,
"codestral-latest": 256000,
"_default": 131072