agent-framework-lib 0.8.18__tar.gz → 0.8.20__tar.gz

{agent_framework_lib-0.8.18/agent_framework_lib.egg-info → agent_framework_lib-0.8.20}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-framework-lib
-Version: 0.8.18
+Version: 0.8.20
 Summary: A comprehensive Python framework for building and serving conversational AI agents with FastAPI
 Author-email: Sebastian Pavel <sebastian@cinco.ai>, Elliott Girard <elliott.girard@icloud.com>
 Maintainer-email: Sebastian Pavel <sebastian@cinco.ai>
@@ -316,7 +316,7 @@ By default, the helper agent uses Claude (if `ANTHROPIC_API_KEY` is set) or GPT-
 
 ```env
 # Force a specific model (useful if your Anthropic key has reached its limit)
-HELPER_AGENT_MODEL=gpt-5.4
+HELPER_AGENT_MODEL=gpt-6-astra
 ```
 
 **Example questions:**
@@ -334,7 +334,7 @@ Pour **déployer** un agent, partez de l'image officielle — Chromium Playwrigh
 sandbox locale déjà en place (sinon l'agent ne peut ni rendre un PNG ni installer quoi que ce soit) :
 
 ```dockerfile
-FROM ghcr.io/komus-ai/agent-framework-runtime:0.8.18
+FROM ghcr.io/komus-ai/agent-framework-runtime:0.8.20
 ```
 
 Détails, publication et visibilité GHCR : [docs/AGENT_RUNTIME_IMAGE.md](docs/AGENT_RUNTIME_IMAGE.md).
@@ -444,14 +444,14 @@ ANTHROPIC_API_KEY=sk-ant-your-anthropic-key
 GEMINI_API_KEY=your-gemini-key
 
 # Model Configuration
-DEFAULT_MODEL=claude-haiku-4-5-20251001
+DEFAULT_MODEL=claude-haiku-5-5
 
 # Multi-Model Routing (Auto Mode)
 DEFAULT_MODEL_MODE=auto                    # "auto" or specific model name
-AUTO_CLASSIFIER_MODEL=gpt-5.4-nano           # Model for complexity classification
-PREFERRED_LIGHT_MODELS=gpt-5.4-nano,claude-haiku-4-5-20251001
-PREFERRED_STANDARD_MODELS=gpt-5.4-mini
-PREFERRED_ADVANCED_MODELS=gpt-5.2,claude-opus-4-6
+AUTO_CLASSIFIER_MODEL=gpt-6-luna           # Model for complexity classification
+PREFERRED_LIGHT_MODELS=gpt-6-luna,claude-haiku-5-5
+PREFERRED_STANDARD_MODELS=gpt-6.1-sol
+PREFERRED_ADVANCED_MODELS=gpt-6-astra,claude-opus-5-5
 
 # Session Storage (optional)
 SESSION_STORAGE_TYPE=memory  # or "mongodb" or "elasticsearch"
@@ -517,9 +517,9 @@ When `DEFAULT_MODEL_MODE=auto`, the system analyzes each query and routes it to
 
 | Tier | Icon | Use Case | Example Models |
 |------|------|----------|----------------|
-| **Light** | 💨 | Simple queries, greetings, basic info | gpt-5.4-nano, claude-haiku-4-5 |
-| **Standard** | ⚖️ | Typical questions, explanations | gpt-5.4-mini, claude-sonnet-4-6 |
-| **Advanced** | 🧠 | Complex analysis, creative tasks | gpt-5.2, claude-opus-4-6 |
+| **Light** | 💨 | Simple queries, greetings, basic info | gpt-6-luna, claude-haiku-5-5 |
+| **Standard** | ⚖️ | Typical questions, explanations | gpt-6.1-sol, claude-sonnet-5-5 |
+| **Advanced** | 🧠 | Complex analysis, creative tasks | gpt-6-astra, claude-opus-5-5 |
 
 **Benefits:**
 - 💰 **Cost optimization** - Use cheaper models for simple queries
@@ -540,12 +540,12 @@ Users can also select a specific model from the UI dropdown:
 DEFAULT_MODEL_MODE=auto
 
 # Model used for complexity classification (should be fast and cheap)
-AUTO_CLASSIFIER_MODEL=gpt-5.4-nano
+AUTO_CLASSIFIER_MODEL=gpt-6-luna
 
 # Preferred models per tier (comma-separated, in order of preference)
-PREFERRED_LIGHT_MODELS=gpt-5.4-nano,claude-haiku-4-5-20251001,gemini-3.5-flash-lite
-PREFERRED_STANDARD_MODELS=gpt-5.4-mini,gemini-3.6-flash
-PREFERRED_ADVANCED_MODELS=gpt-5.2,claude-opus-4-6,gemini-3.1-pro-preview
+PREFERRED_LIGHT_MODELS=gpt-6-luna,claude-haiku-5-5,gemini-3.5-flash-lite
+PREFERRED_STANDARD_MODELS=gpt-6.1-sol,gemini-3.8-flash
+PREFERRED_ADVANCED_MODELS=gpt-6-astra,claude-opus-5-5,gemini-3.1-pro-preview
 ```
 
 ### API Endpoint
@@ -557,12 +557,12 @@ curl http://localhost:8000/api/models
 # Response
 {
   "models_by_tier": {
-    "light": [{"id": "gpt-5.4-nano", "provider": "openai", "available": true}, ...],
+    "light": [{"id": "gpt-6-luna", "provider": "openai", "available": true}, ...],
     "standard": [...],
     "advanced": [...]
   },
   "default_mode": "auto",
-  "classifier_model": "gpt-5.4-nano"
+  "classifier_model": "gpt-6-luna"
 }
 ```
 
@@ -574,7 +574,7 @@ Agents with hardcoded models continue to work without changes:
 class MyAgent(LlamaIndexAgent):
     def __init__(self):
         super().__init__(...)
-        self._default_model = "gpt-5.2"  # This model will always be used
+        self._default_model = "gpt-6-astra"  # This model will always be used
 ```
 
 ### LlamaIndex Agent Configuration
@@ -591,13 +591,13 @@ class MyAgent(LlamaIndexAgent):
         )
         # Default model config (can be overridden per session)
         self.default_temperature = 0.7
-        self.default_model = "claude-haiku-4-5-20251001"
+        self.default_model = "claude-haiku-5-5"
 ```
 
 **Runtime Configuration:**
 
 Users can override settings per session via the API or web UI:
-- Model selection (gpt-5.2, claude-sonnet-4-6, gemini-3.1-pro-preview)
+- Model selection (gpt-6-astra, claude-sonnet-5-5, gemini-3.1-pro-preview)
 - Temperature (0.0 - 1.0)
 - Max tokens
 - System prompt override

{agent_framework_lib-0.8.18/agent_framework/web/docs → agent_framework_lib-0.8.20}/README.md

@@ -155,7 +155,7 @@ By default, the helper agent uses Claude (if `ANTHROPIC_API_KEY` is set) or GPT-
 
 ```env
 # Force a specific model (useful if your Anthropic key has reached its limit)
-HELPER_AGENT_MODEL=gpt-5.4
+HELPER_AGENT_MODEL=gpt-6-astra
 ```
 
 **Example questions:**
@@ -173,7 +173,7 @@ Pour **déployer** un agent, partez de l'image officielle — Chromium Playwrigh
 sandbox locale déjà en place (sinon l'agent ne peut ni rendre un PNG ni installer quoi que ce soit) :
 
 ```dockerfile
-FROM ghcr.io/komus-ai/agent-framework-runtime:0.8.18
+FROM ghcr.io/komus-ai/agent-framework-runtime:0.8.20
 ```
 
 Détails, publication et visibilité GHCR : [docs/AGENT_RUNTIME_IMAGE.md](docs/AGENT_RUNTIME_IMAGE.md).
@@ -283,14 +283,14 @@ ANTHROPIC_API_KEY=sk-ant-your-anthropic-key
 GEMINI_API_KEY=your-gemini-key
 
 # Model Configuration
-DEFAULT_MODEL=claude-haiku-4-5-20251001
+DEFAULT_MODEL=claude-haiku-5-5
 
 # Multi-Model Routing (Auto Mode)
 DEFAULT_MODEL_MODE=auto                    # "auto" or specific model name
-AUTO_CLASSIFIER_MODEL=gpt-5.4-nano           # Model for complexity classification
-PREFERRED_LIGHT_MODELS=gpt-5.4-nano,claude-haiku-4-5-20251001
-PREFERRED_STANDARD_MODELS=gpt-5.4-mini
-PREFERRED_ADVANCED_MODELS=gpt-5.2,claude-opus-4-6
+AUTO_CLASSIFIER_MODEL=gpt-6-luna           # Model for complexity classification
+PREFERRED_LIGHT_MODELS=gpt-6-luna,claude-haiku-5-5
+PREFERRED_STANDARD_MODELS=gpt-6.1-sol
+PREFERRED_ADVANCED_MODELS=gpt-6-astra,claude-opus-5-5
 
 # Session Storage (optional)
 SESSION_STORAGE_TYPE=memory  # or "mongodb" or "elasticsearch"
@@ -356,9 +356,9 @@ When `DEFAULT_MODEL_MODE=auto`, the system analyzes each query and routes it to
 
 | Tier | Icon | Use Case | Example Models |
 |------|------|----------|----------------|
-| **Light** | 💨 | Simple queries, greetings, basic info | gpt-5.4-nano, claude-haiku-4-5 |
-| **Standard** | ⚖️ | Typical questions, explanations | gpt-5.4-mini, claude-sonnet-4-6 |
-| **Advanced** | 🧠 | Complex analysis, creative tasks | gpt-5.2, claude-opus-4-6 |
+| **Light** | 💨 | Simple queries, greetings, basic info | gpt-6-luna, claude-haiku-5-5 |
+| **Standard** | ⚖️ | Typical questions, explanations | gpt-6.1-sol, claude-sonnet-5-5 |
+| **Advanced** | 🧠 | Complex analysis, creative tasks | gpt-6-astra, claude-opus-5-5 |
 
 **Benefits:**
 - 💰 **Cost optimization** - Use cheaper models for simple queries
@@ -379,12 +379,12 @@ Users can also select a specific model from the UI dropdown:
 DEFAULT_MODEL_MODE=auto
 
 # Model used for complexity classification (should be fast and cheap)
-AUTO_CLASSIFIER_MODEL=gpt-5.4-nano
+AUTO_CLASSIFIER_MODEL=gpt-6-luna
 
 # Preferred models per tier (comma-separated, in order of preference)
-PREFERRED_LIGHT_MODELS=gpt-5.4-nano,claude-haiku-4-5-20251001,gemini-3.5-flash-lite
-PREFERRED_STANDARD_MODELS=gpt-5.4-mini,gemini-3.6-flash
-PREFERRED_ADVANCED_MODELS=gpt-5.2,claude-opus-4-6,gemini-3.1-pro-preview
+PREFERRED_LIGHT_MODELS=gpt-6-luna,claude-haiku-5-5,gemini-3.5-flash-lite
+PREFERRED_STANDARD_MODELS=gpt-6.1-sol,gemini-3.8-flash
+PREFERRED_ADVANCED_MODELS=gpt-6-astra,claude-opus-5-5,gemini-3.1-pro-preview
 ```
 
 ### API Endpoint
@@ -396,12 +396,12 @@ curl http://localhost:8000/api/models
 # Response
 {
   "models_by_tier": {
-    "light": [{"id": "gpt-5.4-nano", "provider": "openai", "available": true}, ...],
+    "light": [{"id": "gpt-6-luna", "provider": "openai", "available": true}, ...],
     "standard": [...],
     "advanced": [...]
   },
   "default_mode": "auto",
-  "classifier_model": "gpt-5.4-nano"
+  "classifier_model": "gpt-6-luna"
 }
 ```
 
@@ -413,7 +413,7 @@ Agents with hardcoded models continue to work without changes:
 class MyAgent(LlamaIndexAgent):
     def __init__(self):
         super().__init__(...)
-        self._default_model = "gpt-5.2"  # This model will always be used
+        self._default_model = "gpt-6-astra"  # This model will always be used
 ```
 
 ### LlamaIndex Agent Configuration
@@ -430,13 +430,13 @@ class MyAgent(LlamaIndexAgent):
         )
         # Default model config (can be overridden per session)
         self.default_temperature = 0.7
-        self.default_model = "claude-haiku-4-5-20251001"
+        self.default_model = "claude-haiku-5-5"
 ```
 
 **Runtime Configuration:**
 
 Users can override settings per session via the API or web UI:
-- Model selection (gpt-5.2, claude-sonnet-4-6, gemini-3.1-pro-preview)
+- Model selection (gpt-6-astra, claude-sonnet-5-5, gemini-3.1-pro-preview)
 - Temperature (0.0 - 1.0)
 - Max tokens
 - System prompt override

{agent_framework_lib-0.8.18 → agent_framework_lib-0.8.20}/agent_framework/core/base_agent.py

@@ -89,8 +89,18 @@ logger = logging.getLogger(__name__)
 DEFAULT_AGENT_TIMEZONE = "Europe/Paris"
 _JOURS = ("lundi", "mardi", "mercredi", "jeudi", "vendredi", "samedi", "dimanche")
 _MOIS = (
-    "janvier", "février", "mars", "avril", "mai", "juin",
-    "juillet", "août", "septembre", "octobre", "novembre", "décembre",
+    "janvier",
+    "février",
+    "mars",
+    "avril",
+    "mai",
+    "juin",
+    "juillet",
+    "août",
+    "septembre",
+    "octobre",
+    "novembre",
+    "décembre",
 )
 
 
@@ -977,7 +987,8 @@ class BaseAgent(SkillsMixin, MemoryMixin, AgentInterface):
     # ----- AgentInterface -----
     async def configure_session(self, session_configuration: dict[str, Any]) -> None:
         """Configure the agent with session-level settings."""
-        logger.info(f"BaseAgent: Configuring session: {session_configuration}")
+        # Key names only: values can carry client secrets (portal_api_key, tools).
+        logger.info(f"BaseAgent: Configuring session with keys: {sorted(session_configuration)}")
 
         # Handle skills configuration (replaces rich content configuration)
         if "enable_skills" in session_configuration:

{agent_framework_lib-0.8.18 → agent_framework_lib-0.8.20}/agent_framework/core/context_budget.py

@@ -39,6 +39,7 @@ class ModelContextRegistry:
         # Standard tier
         "gpt-5.4-mini": 400_000,
         "claude-haiku-4-5-20251001": 200_000,
+        "claude-haiku-5-5": 1_000_000,
         "gemini-2.5-flash": 1_048_576,
         "gemini-3-flash-preview": 1_048_576,
         "gemini-3.5-flash": 1_048_576,
@@ -70,6 +71,7 @@ class ModelContextRegistry:
         "gpt-6-luna": 1_050_000,
         "gpt-6-sol": 1_050_000,
         "gpt-6-astra": 1_050_000,
+        "gpt-6.1-sol": 1_050_000,
         "gemini-2.5-pro": 1_048_576,
         "gemini-3.1-pro-preview": 1_048_576,
         # Mistral — valeurs max_context_length de GET /v1/models, compte Komus, 17/09/2026.

{agent_framework_lib-0.8.18 → agent_framework_lib-0.8.20}/agent_framework/core/model_config.py

@@ -17,7 +17,7 @@ Environment Variables:
 - GEMINI_API_KEY: Google Gemini API key
 - MISTRAL_API_KEY: Mistral AI API key
 - MISTRAL_BASE_URL: OpenAI-compatible endpoint for Mistral (default: https://api.mistral.ai/v1)
-- DEFAULT_MODEL: Default model to use (default: "claude-haiku-4-5-20251001")
+- DEFAULT_MODEL: Default model to use (default: "claude-haiku-5-5")
 - OPENAI_MODELS / ANTHROPIC_MODELS / GEMINI_MODELS / MISTRAL_MODELS: Comma-separated
   lists of known model names per provider (override the built-in lists)
 - FALLBACK_PROVIDER: Provider to use when model provider is unknown
@@ -57,7 +57,7 @@ logger = logging.getLogger(__name__)
 # Modèle par défaut du framework quand rien n'est configuré (DEFAULT_MODEL absent,
 # classifieur du routeur, titres de session, compresseur, agent d'aide…).
 # Décision Komus 15/09/2026 : Claude Haiku 4.5 partout (rapide, peu cher, vision).
-FRAMEWORK_DEFAULT_MODEL: Final[str] = "claude-haiku-4-5-20251001"
+FRAMEWORK_DEFAULT_MODEL: Final[str] = "claude-haiku-5-5"
 
 # Endpoint OpenAI-compatible de Mistral (« La Plateforme »). Surchargeable via
 # MISTRAL_BASE_URL (proxy d'entreprise, vLLM…).
@@ -113,52 +113,29 @@ class ModelConfigManager:
         gemini_defaults: Default parameters for Gemini models
     """
 
-    # Default model mappings (can be overridden by environment variables)
-    # Modèles constatés sur les comptes Komus le 15/09/2026, revus le 29/09/2026 : seuls
-    # les ids appelables à travers la lib (chat + stream + outil) sont proposés. Un id absent
-    # d'ici reste routé par motif (« gpt », « claude », « gemini »). Les modèles de
-    # ``OPENAI_RESPONSES_API_MODELS`` passent par l'API Responses (voir plus bas).
+    # Listes proposées : le DERNIER modèle de chaque gamme seulement (09/10/2026, demande
+    # d'Elliott : « la liste est trop longue »). Un id retiré reste utilisable s'il est demandé
+    # explicitement : il est routé vers son fournisseur par motif (« gpt », « claude »,
+    # « gemini »). Tous sondés à travers la lib (chat + stream + outil) le 09/10/2026.
+    # OpenAI : GPT-6 Luna / Sol / Astra (léger / standard / avancé), servis par l'API
+    # Responses (voir OPENAI_RESPONSES_API_MODELS).
     DEFAULT_OPENAI_MODELS: Final[list[str]] = [
-        "gpt-5.4-nano",
-        "gpt-5.4-mini",
-        "gpt-5.4",
-        "gpt-5.4-pro",
-        "gpt-5.5",
-        "gpt-5.5-pro",
-        "gpt-5.6-sol",
-        "gpt-5.6-terra",
-        "gpt-5.6-luna",
         "gpt-6-luna",
-        "gpt-6-sol",
+        "gpt-6.1-sol",
         "gpt-6-astra",
     ]
 
     DEFAULT_ANTHROPIC_MODELS: Final[list[str]] = [
-        "claude-haiku-4-5-20251001",
-        "claude-sonnet-4-6",
-        "claude-opus-4-5-20251101",
-        "claude-opus-4-6",
-        "claude-opus-4-7",
-        "claude-opus-4-8",
-        "claude-sonnet-5",
-        "claude-opus-5",
+        "claude-haiku-5-5",
         "claude-sonnet-5-5",
         "claude-opus-5-5",
-        "claude-fable-5",
         "claude-fable-5-1",
     ]
 
-    # Gemini sondés le 29/09/2026 à travers la lib (chat + stream + outil, tous OK).
-    # gemini-3.1-flash-lite-preview est arrêté ; la famille 2.5 répond 404 « no longer
-    # available to new users » sur le compte Komus.
+    # Gemini : un modèle par gamme (Flash-Lite, Flash, Pro).
     DEFAULT_GEMINI_MODELS: Final[list[str]] = [
-        "gemini-3.1-flash-lite",
         "gemini-3.5-flash-lite",
-        "gemini-3.5-flash",
-        "gemini-3.6-flash",
-        "gemini-3.7-flash",
         "gemini-3.8-flash",
-        "gemini-3-flash-preview",
         "gemini-3.1-pro-preview",
     ]
 
@@ -176,13 +153,10 @@ class ModelConfigManager:
     # Hors liste par défaut (toujours routés par motif si on les nomme) :
     # - mistral-small/medium-latest, magistral-*, devstral-* : 429 quota 0 hors plan ;
     # - mistral-large-latest : 403 « not available in your subscription tier ».
+    # 09/10/2026 : seuls les deux modèles des tiers sont proposés (léger, standard/avancé).
     DEFAULT_MISTRAL_MODELS: Final[list[str]] = [
-        "ministral-3b-latest",
         "ministral-8b-latest",
-        "open-mistral-nemo",
-        "pixtral-12b-latest",
         "ministral-14b-latest",
-        "codestral-latest",
     ]
 
     def __init__(self) -> None:
@@ -674,7 +648,7 @@ class ModelConfigManager:
             ```python
             model_config.register_agent_config("my-agent", {
                 "system_prompt": "You are a helpful assistant.",
-                "model_name": "claude-haiku-4-5-20251001",
+                "model_name": "claude-haiku-5-5",
                 "model_config": {"temperature": 0.7, "max_tokens": 50000}
             })
             ```
@@ -1036,6 +1010,7 @@ ANTHROPIC_NO_SAMPLING_MODELS: Final[tuple[str, ...]] = (
     "claude-opus-5",
     "claude-sonnet-5-5",
     "claude-opus-5-5",
+    "claude-haiku-5",
     "claude-fable-5",
 )
 

{agent_framework_lib-0.8.18 → agent_framework_lib-0.8.20}/agent_framework/core/model_router.py

@@ -134,36 +134,24 @@ class ModelRouter:
     # Mistral en repli de fin de liste (intégré le 17/09/2026) : jamais devant Claude.
     # 29/09/2026 : OpenAI vérifié à travers la lib (chat + stream + outil) ; Gemini sondé de même
     # (clé de l'env dev Mozart) ; gpt-5.5-pro sorti (Responses API seulement).
+    # 09/10/2026 : un seul modèle par fournisseur et par tier, le dernier de chaque gamme ;
+    # Claude Haiku 5.5 est le modèle de base du mode auto (tête du tier léger).
     DEFAULT_LIGHT_MODELS: list[str] = [
-        "claude-haiku-4-5-20251001",
-        "gpt-5.4-nano",
+        "claude-haiku-5-5",
+        "gpt-6-luna",
         "gemini-3.5-flash-lite",
-        "gemini-3.1-flash-lite",
         "ministral-8b-latest",
     ]
     DEFAULT_STANDARD_MODELS: list[str] = [
         "claude-sonnet-5-5",
-        "claude-sonnet-5",
-        "claude-sonnet-4-6",
-        "gpt-5.4-mini",
-        "gemini-3.6-flash",
-        "gemini-3.5-flash",
+        "gpt-6.1-sol",
+        "gemini-3.8-flash",
         "ministral-14b-latest",
     ]
     DEFAULT_ADVANCED_MODELS: list[str] = [
         "claude-opus-5-5",
-        "claude-opus-5",
         "claude-fable-5-1",
-        "claude-opus-4-8",
-        "gpt-5.5",
-        "claude-sonnet-5-5",
-        "claude-sonnet-5",
-        "claude-sonnet-4-6",
-        "claude-opus-4-6",
-        "claude-opus-4-5-20251101",
-        "gpt-5.4",
-        "gemini-3.8-flash",
-        "gemini-3.7-flash",
+        "gpt-6-astra",
         "gemini-3.1-pro-preview",
         # Un modèle est jugé disponible dès que la clé du provider existe — le routeur ne
         # sait pas si le PLAN du compte le sert (429 quota 0 / 403 sinon, sans repli). Seuls

{agent_framework_lib-0.8.18 → agent_framework_lib-0.8.20}/agent_framework/memory/providers/memori_provider.py

@@ -40,6 +40,7 @@ Example:
 Version: 0.1.0
 """
 
+import importlib.util
 import logging
 from datetime import datetime
 from typing import Any
@@ -50,17 +51,10 @@ from ..config import MemoriConfig
 
 logger = logging.getLogger(__name__)
 
-# Lazy import to avoid dependency issues if Memori not installed
-MEMORI_AVAILABLE = False
-_memori_import_error: str | None = None
-
-try:
-    from memori import Memori
-
-    MEMORI_AVAILABLE = True
-except ImportError as e:
-    Memori = None
-    _memori_import_error = str(e)
+# Availability only: importing memori pulls sentence_transformers/transformers
+# (~8 s), so the import itself waits for initialize().
+MEMORI_AVAILABLE = importlib.util.find_spec("memori") is not None
+_memori_import_error: str | None = None if MEMORI_AVAILABLE else "No module named 'memori'"
 
 
 class MemoriProvider(MemoryProviderInterface):
@@ -118,6 +112,7 @@ class MemoriProvider(MemoryProviderInterface):
             return True
 
         try:
+            from memori import Memori
             from sqlalchemy import create_engine
             from sqlalchemy.orm import sessionmaker
 

{agent_framework_lib-0.8.18 → agent_framework_lib-0.8.20}/agent_framework/utils/session_title_generator.py

@@ -12,7 +12,7 @@ Key Features:
 
 Environment Variables:
 - AUTO_GENERATE_SESSION_TITLES: Enable/disable auto-generation (default: "true")
-- SESSION_TITLE_MODEL: LLM model for generation (default: "claude-haiku-4-5-20251001")
+- SESSION_TITLE_MODEL: LLM model for generation (default: "claude-haiku-5-5")
 - SESSION_TITLE_MAX_TOKENS: Max tokens for title (default: 20)
 - SESSION_TITLE_TEMPERATURE: LLM temperature (default: 0.3)
 """

{agent_framework_lib-0.8.18 → agent_framework_lib-0.8.20}/agent_framework/web/docs/Dockerfile

@@ -2,7 +2,7 @@
 # Base = image de runtime officielle (Chromium Playwright, Deno, git, root, sandbox locale) :
 # voir docs/AGENT_RUNTIME_IMAGE.md. Ne repartez PAS de python:*-slim à la main : vous
 # retomberiez sur une sandbox sans libs Chromium ni droits (export PNG impossible).
-FROM ghcr.io/komus-ai/agent-framework-runtime:0.8.18
+FROM ghcr.io/komus-ai/agent-framework-runtime:0.8.20
 
 WORKDIR /app
 

{agent_framework_lib-0.8.18 → agent_framework_lib-0.8.20/agent_framework/web/docs}/README.md

@@ -155,7 +155,7 @@ By default, the helper agent uses Claude (if `ANTHROPIC_API_KEY` is set) or GPT-
 
 ```env
 # Force a specific model (useful if your Anthropic key has reached its limit)
-HELPER_AGENT_MODEL=gpt-5.4
+HELPER_AGENT_MODEL=gpt-6-astra
 ```
 
 **Example questions:**
@@ -173,7 +173,7 @@ Pour **déployer** un agent, partez de l'image officielle — Chromium Playwrigh
 sandbox locale déjà en place (sinon l'agent ne peut ni rendre un PNG ni installer quoi que ce soit) :
 
 ```dockerfile
-FROM ghcr.io/komus-ai/agent-framework-runtime:0.8.18
+FROM ghcr.io/komus-ai/agent-framework-runtime:0.8.20
 ```
 
 Détails, publication et visibilité GHCR : [docs/AGENT_RUNTIME_IMAGE.md](docs/AGENT_RUNTIME_IMAGE.md).
@@ -283,14 +283,14 @@ ANTHROPIC_API_KEY=sk-ant-your-anthropic-key
 GEMINI_API_KEY=your-gemini-key
 
 # Model Configuration
-DEFAULT_MODEL=claude-haiku-4-5-20251001
+DEFAULT_MODEL=claude-haiku-5-5
 
 # Multi-Model Routing (Auto Mode)
 DEFAULT_MODEL_MODE=auto                    # "auto" or specific model name
-AUTO_CLASSIFIER_MODEL=gpt-5.4-nano           # Model for complexity classification
-PREFERRED_LIGHT_MODELS=gpt-5.4-nano,claude-haiku-4-5-20251001
-PREFERRED_STANDARD_MODELS=gpt-5.4-mini
-PREFERRED_ADVANCED_MODELS=gpt-5.2,claude-opus-4-6
+AUTO_CLASSIFIER_MODEL=gpt-6-luna           # Model for complexity classification
+PREFERRED_LIGHT_MODELS=gpt-6-luna,claude-haiku-5-5
+PREFERRED_STANDARD_MODELS=gpt-6.1-sol
+PREFERRED_ADVANCED_MODELS=gpt-6-astra,claude-opus-5-5
 
 # Session Storage (optional)
 SESSION_STORAGE_TYPE=memory  # or "mongodb" or "elasticsearch"
@@ -356,9 +356,9 @@ When `DEFAULT_MODEL_MODE=auto`, the system analyzes each query and routes it to
 
 | Tier | Icon | Use Case | Example Models |
 |------|------|----------|----------------|
-| **Light** | 💨 | Simple queries, greetings, basic info | gpt-5.4-nano, claude-haiku-4-5 |
-| **Standard** | ⚖️ | Typical questions, explanations | gpt-5.4-mini, claude-sonnet-4-6 |
-| **Advanced** | 🧠 | Complex analysis, creative tasks | gpt-5.2, claude-opus-4-6 |
+| **Light** | 💨 | Simple queries, greetings, basic info | gpt-6-luna, claude-haiku-5-5 |
+| **Standard** | ⚖️ | Typical questions, explanations | gpt-6.1-sol, claude-sonnet-5-5 |
+| **Advanced** | 🧠 | Complex analysis, creative tasks | gpt-6-astra, claude-opus-5-5 |
 
 **Benefits:**
 - 💰 **Cost optimization** - Use cheaper models for simple queries
@@ -379,12 +379,12 @@ Users can also select a specific model from the UI dropdown:
 DEFAULT_MODEL_MODE=auto
 
 # Model used for complexity classification (should be fast and cheap)
-AUTO_CLASSIFIER_MODEL=gpt-5.4-nano
+AUTO_CLASSIFIER_MODEL=gpt-6-luna
 
 # Preferred models per tier (comma-separated, in order of preference)
-PREFERRED_LIGHT_MODELS=gpt-5.4-nano,claude-haiku-4-5-20251001,gemini-3.5-flash-lite
-PREFERRED_STANDARD_MODELS=gpt-5.4-mini,gemini-3.6-flash
-PREFERRED_ADVANCED_MODELS=gpt-5.2,claude-opus-4-6,gemini-3.1-pro-preview
+PREFERRED_LIGHT_MODELS=gpt-6-luna,claude-haiku-5-5,gemini-3.5-flash-lite
+PREFERRED_STANDARD_MODELS=gpt-6.1-sol,gemini-3.8-flash
+PREFERRED_ADVANCED_MODELS=gpt-6-astra,claude-opus-5-5,gemini-3.1-pro-preview
 ```
 
 ### API Endpoint
@@ -396,12 +396,12 @@ curl http://localhost:8000/api/models
 # Response
 {
   "models_by_tier": {
-    "light": [{"id": "gpt-5.4-nano", "provider": "openai", "available": true}, ...],
+    "light": [{"id": "gpt-6-luna", "provider": "openai", "available": true}, ...],
     "standard": [...],
     "advanced": [...]
   },
   "default_mode": "auto",
-  "classifier_model": "gpt-5.4-nano"
+  "classifier_model": "gpt-6-luna"
 }
 ```
 
@@ -413,7 +413,7 @@ Agents with hardcoded models continue to work without changes:
 class MyAgent(LlamaIndexAgent):
     def __init__(self):
         super().__init__(...)
-        self._default_model = "gpt-5.2"  # This model will always be used
+        self._default_model = "gpt-6-astra"  # This model will always be used
 ```
 
 ### LlamaIndex Agent Configuration
@@ -430,13 +430,13 @@ class MyAgent(LlamaIndexAgent):
         )
         # Default model config (can be overridden per session)
         self.default_temperature = 0.7
-        self.default_model = "claude-haiku-4-5-20251001"
+        self.default_model = "claude-haiku-5-5"
 ```
 
 **Runtime Configuration:**
 
 Users can override settings per session via the API or web UI:
-- Model selection (gpt-5.2, claude-sonnet-4-6, gemini-3.1-pro-preview)
+- Model selection (gpt-6-astra, claude-sonnet-5-5, gemini-3.1-pro-preview)
 - Temperature (0.0 - 1.0)
 - Max tokens
 - System prompt override