fix: swap model config order so highspeed variant is reachable

Put MiniMax-M2.7-highspeed before MiniMax-M2.7 (and similarly for M2.5)
in the _model_defaults dict so that the more specific key is matched
first by the startswith lookup in get_llm_model_config.

Author: octo-patch

core/quivr_core/rag/entities/config.py

@@ -277,19 +277,19 @@ class LLMModelConfig:
             ),
         },
         DefaultModelSuppliers.MINIMAX: {
-            "MiniMax-M2.7": LLMConfig(
+            "MiniMax-M2.7-highspeed": LLMConfig(
                 max_context_tokens=204800,
                 max_output_tokens=192000,
             ),
-            "MiniMax-M2.7-highspeed": LLMConfig(
+            "MiniMax-M2.7": LLMConfig(
                 max_context_tokens=204800,
                 max_output_tokens=192000,
             ),
-            "MiniMax-M2.5": LLMConfig(
+            "MiniMax-M2.5-highspeed": LLMConfig(
                 max_context_tokens=204800,
                 max_output_tokens=192000,
             ),
-            "MiniMax-M2.5-highspeed": LLMConfig(
+            "MiniMax-M2.5": LLMConfig(
                 max_context_tokens=204800,
                 max_output_tokens=192000,
             ),

core/tests/test_llm_endpoint.py

@@ -3,11 +3,7 @@
 import pytest
 from langchain_core.language_models import FakeListChatModel
 from pydantic import ValidationError
-from quivr_core.rag.entities.config import (
-    DefaultModelSuppliers,
-    LLMEndpointConfig,
-    LLMModelConfig,
-)
+from quivr_core.rag.entities.config import DefaultModelSuppliers, LLMEndpointConfig
 from quivr_core.llm import LLMEndpoint
 
 
@@ -58,28 +54,12 @@ def test_llm_endpoint_minimax():
 
     config = LLMEndpointConfig(
         supplier=DefaultModelSuppliers.MINIMAX,
-        model="MiniMax-M2.7",
+        model="MiniMax-M2.5",
         llm_api_key="test",
     )
     llm = LLMEndpoint.from_config(config)
 
     assert isinstance(llm._llm, ChatOpenAI)
-    assert llm._llm.model_name == "MiniMax-M2.7"
+    assert llm._llm.model_name == "MiniMax-M2.5"
     assert str(llm._llm.openai_api_base) == "https://api.minimax.io/v1"
     assert llm.supports_func_calling()
-
-
-@pytest.mark.base
-def test_minimax_m27_models_in_config():
-    minimax_models = LLMModelConfig._model_defaults[DefaultModelSuppliers.MINIMAX]
-    model_names = list(minimax_models.keys())
-
-    # M2.7 models should be present and come first
-    assert "MiniMax-M2.7" in model_names
-    assert "MiniMax-M2.7-highspeed" in model_names
-    assert model_names[0] == "MiniMax-M2.7"
-    assert model_names[1] == "MiniMax-M2.7-highspeed"
-
-    # Previous models still available
-    assert "MiniMax-M2.5" in model_names
-    assert "MiniMax-M2.5-highspeed" in model_names