diff --git a/.env.example b/.env.example
index fa458d74..07ccf0a3 100644
--- a/.env.example
+++ b/.env.example
@@ -32,26 +32,4 @@ FAST_PREFIX_DETECTION=true
 ENABLE_NETWORK_PROBE_MOCK=true
 ENABLE_TITLE_GENERATION_SKIP=true
 ENABLE_SUGGESTION_MODE_SKIP=true
-ENABLE_FILEPATH_EXTRACTION_MOCK=true
-
-
-# All NVIDIA_NIM_* settings are strictly validated (unknown keys will error).
-NVIDIA_NIM_TEMPERATURE=1.0
-NVIDIA_NIM_TOP_P=1.0
-NVIDIA_NIM_TOP_K=-1
-NVIDIA_NIM_MAX_TOKENS=81920
-NVIDIA_NIM_PRESENCE_PENALTY=0.0
-NVIDIA_NIM_FREQUENCY_PENALTY=0.0
-NVIDIA_NIM_MIN_P=0.0
-NVIDIA_NIM_REPETITION_PENALTY=1.0
-NVIDIA_NIM_SEED=
-NVIDIA_NIM_STOP=
-NVIDIA_NIM_PARALLEL_TOOL_CALLS=true
-NVIDIA_NIM_RETURN_TOKENS_AS_TOKEN_IDS=false
-NVIDIA_NIM_INCLUDE_STOP_STR_IN_OUTPUT=false
-NVIDIA_NIM_IGNORE_EOS=false
-NVIDIA_NIM_MIN_TOKENS=0
-NVIDIA_NIM_CHAT_TEMPLATE=""
-NVIDIA_NIM_REQUEST_ID=""
-NVIDIA_NIM_REASONING_EFFORT=high
-NVIDIA_NIM_INCLUDE_REASONING=true
\ No newline at end of file
+ENABLE_FILEPATH_EXTRACTION_MOCK=true
\ No newline at end of file
diff --git a/README.md b/README.md
index 1783077c..6dfc04b1 100644
--- a/README.md
+++ b/README.md
@@ -219,32 +219,6 @@ Browse: [openrouter.ai/models](https://openrouter.ai/models)
 - **NVIDIA NIM** base URL: `https://integrate.api.nvidia.com/v1`
 - **OpenRouter** base URL: `https://openrouter.ai/api/v1`
 
-**NIM Settings (prefix `NVIDIA_NIM_`)**
-
-| Variable                                | Description                   | Default |
-| --------------------------------------- | ----------------------------- | ------- |
-| `NVIDIA_NIM_TEMPERATURE`                | Sampling temperature          | `1.0`   |
-| `NVIDIA_NIM_TOP_P`                      | Top-p nucleus sampling        | `1.0`   |
-| `NVIDIA_NIM_TOP_K`                      | Top-k sampling                | `-1`    |
-| `NVIDIA_NIM_MAX_TOKENS`                 | Max tokens for generation     | `81920` |
-| `NVIDIA_NIM_PRESENCE_PENALTY`           | Presence penalty              | `0.0`   |
-| `NVIDIA_NIM_FREQUENCY_PENALTY`          | Frequency penalty             | `0.0`   |
-| `NVIDIA_NIM_MIN_P`                      | Min-p sampling                | `0.0`   |
-| `NVIDIA_NIM_REPETITION_PENALTY`         | Repetition penalty            | `1.0`   |
-| `NVIDIA_NIM_SEED`                       | RNG seed (blank = unset)      | unset   |
-| `NVIDIA_NIM_STOP`                       | Stop string (blank = unset)   | unset   |
-| `NVIDIA_NIM_PARALLEL_TOOL_CALLS`        | Parallel tool calls           | `true`  |
-| `NVIDIA_NIM_RETURN_TOKENS_AS_TOKEN_IDS` | Return token ids              | `false` |
-| `NVIDIA_NIM_INCLUDE_STOP_STR_IN_OUTPUT` | Include stop string in output | `false` |
-| `NVIDIA_NIM_IGNORE_EOS`                 | Ignore EOS token              | `false` |
-| `NVIDIA_NIM_MIN_TOKENS`                 | Minimum generated tokens      | `0`     |
-| `NVIDIA_NIM_CHAT_TEMPLATE`              | Chat template override        | unset   |
-| `NVIDIA_NIM_REQUEST_ID`                 | Request id override           | unset   |
-| `NVIDIA_NIM_REASONING_EFFORT`           | Reasoning effort              | `high`  |
-| `NVIDIA_NIM_INCLUDE_REASONING`          | Include reasoning in response | `true`  |
-
-All `NVIDIA_NIM_*` settings are strictly validated; unknown keys with this prefix will cause startup errors.
-
 See [`.env.example`](.env.example) for all supported parameters.
 
 ## Development
diff --git a/config/nim.py b/config/nim.py
index a95af482..83dd3cd2 100644
--- a/config/nim.py
+++ b/config/nim.py
@@ -1,13 +1,12 @@
-"""NVIDIA NIM settings (strict validation)."""
+"""NVIDIA NIM settings (fixed values, no env config)."""
 
-from typing import Optional, Literal
+from typing import Literal, Optional
 
-from pydantic import Field, field_validator
-from pydantic_settings import BaseSettings, SettingsConfigDict
+from pydantic import BaseModel, ConfigDict, Field, field_validator
 
 
-class NimSettings(BaseSettings):
-    """Strictly validated NVIDIA NIM settings."""
+class NimSettings(BaseModel):
+    """Fixed NVIDIA NIM settings (not configurable via env)."""
 
     temperature: float = Field(1.0, ge=0.0)
     top_p: float = Field(1.0, ge=0.0, le=1.0)
@@ -34,12 +33,7 @@ class NimSettings(BaseSettings):
     reasoning_effort: Literal["low", "medium", "high"] = "high"
     include_reasoning: bool = True
 
-    model_config = SettingsConfigDict(
-        env_prefix="NVIDIA_NIM_",
-        # Rely on global load_dotenv in config.settings to avoid
-        # reading unrelated .env keys into this settings model.
-        extra="forbid",
-    )
+    model_config = ConfigDict(extra="forbid")
 
     @field_validator("top_k")
     @classmethod