Skip to content

Add vLLM Semantic Router (vllm-sr) integration #61

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
yl231 merged 8 commits into from
Feb 2, 2026
Merged

Add vLLM Semantic Router (vllm-sr) integration #61

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
ba82450
generate_prediction_file create router instance according to router_n…
carlory Jan 22, 2026
56a3c33
add vllm sr
carlory Jan 22, 2026
3472e47
fix readme
carlory Jan 22, 2026
4a28cf6
Fix ruff formatting in vllm_sr.py
carlory Jan 22, 2026
6a9ee29
update vllm-sr
carlory Jan 23, 2026
29baf7e
uv run python ./llm_inference/run.py vllm-sr
carlory Jan 23, 2026
952ae27
uv run python ./router_inference/generate_prediction_file.py vllm-sr …
carlory Jan 28, 2026
cd9eeb4
Update model_cost.json and vllm-sr predictions
ZhiboYu1 Jan 30, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 5 additions & 3 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -102,6 +102,7 @@ Create a config file in `./router_inference/config/<router_name>.json`. An examp
{
"pipeline_params": {
"router_name": "your-router",
"router_cls_name": "your_router_class_name",
"models": [
"gpt-4o-mini",
"claude-3-haiku-20240307",
Expand Down Expand Up @@ -129,12 +130,13 @@ For each model in your config, add an entry with the pricing per million tokens

Create your own router class by inheriting from `BaseRouter` and implementing the `_get_prediction()` method. See [`router_inference/router/example_router.py`](./router_inference/router/example_router.py) for a complete example.

Then, modify [`router_inference/generate_prediction_file.py`](./router_inference/generate_prediction_file.py#L150) to use your router class:
Then, modify [`router_inference/router/__init__.py`](./router_inference/router/__init__.py) to include your router class:

```python
# Replace ExampleRouter with your router class
# Import your router class
from router_inference.router.my_router import MyRouter
router = MyRouter(args.router_name)

__all__ = ["BaseRouter", "ExampleRouter", "MyRouter"]
```

Finally, generate the prediction file:
Expand Down
332 changes: 90 additions & 242 deletions model_cost/model_cost.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,133 +1,21 @@
{
"Qwen/QwQ-32B": {
"input_token_price_per_million": 1.2,
"output_token_price_per_million": 1.2
},
"anthropic/claude-3.5-sonnet": {
"input_token_price_per_million": 3,
"output_token_price_per_million": 15
},
"claude-3-7-sonnet-20250219": {
"input_token_price_per_million": 3,
"output_token_price_per_million": 15
},
"claude-3-haiku-20240307": {
"input_token_price_per_million": 0.25,
"output_token_price_per_million": 1.25
},
"claude-sonnet-4-5": {
"input_token_price_per_million": 3.0,
"output_token_price_per_million": 15.0
},
"codestral-latest": {
"input_token_price_per_million": 0.3,
"output_token_price_per_million": 0.9
},
"deepseek-coder": {
"input_token_price_per_million": 0.25,
"output_token_price_per_million": 1.0
},
"deepseek-reasoner": {
"input_token_price_per_million": 0.28,
"output_token_price_per_million": 0.42
},
"deepseek-v3.2": {
"input_token_price_per_million": 0.28,
"output_token_price_per_million": 0.42
},
"gemini-2.0-flash-001": {
"input_token_price_per_million": 0.1,
"output_token_price_per_million": 0.4
},
"gemini-2.5-flash": {
"input_token_price_per_million": 0.3,
"output_token_price_per_million": 2.5
},
"gemini-2.5-flash-lite": {
"input_token_price_per_million": 0.1,
"output_token_price_per_million": 0.4
},
"gemini-2.5-pro": {
"input_token_price_per_million": 1.25,
"output_token_price_per_million": 10.0
},
"gemini-3-flash-preview": {
"input_token_price_per_million": 0.5,
"output_token_price_per_million": 3
},
"gemini-3-pro-preview": {
"input_token_price_per_million": 2.0,
"output_token_price_per_million": 12.0
},
"gemini-3.0-pro": {
"input_token_price_per_million": 2.0,
"output_token_price_per_million": 12.0
},
"glm-4-air": {
"input_token_price_per_million": 0.137,
"output_token_price_per_million": 0.137
},
"glm-4-air-250414": {
"input_token_price_per_million": 0.07,
"output_token_price_per_million": 0.07
},
"glm-4-flash": {
"input_token_price_per_million": 0.0137,
"output_token_price_per_million": 0.0137
},
"glm-4-plus": {
"input_token_price_per_million": 0.7,
"output_token_price_per_million": 0.7
},
"glm-4.5-air": {
"input_token_price_per_million": 0.1,
"output_token_price_per_million": 0.3
},
"glm-4.6": {
"input_token_price_per_million": 0.28,
"output_token_price_per_million": 1.12
},
"gpt-3.5-turbo": {
"input_token_price_per_million": 1,
"output_token_price_per_million": 2
},
"gpt-3.5-turbo-1106": {
"input_token_price_per_million": 1,
"output_token_price_per_million": 2
},
"gpt-4": {
"input_token_price_per_million": 10,
"output_token_price_per_million": 30
},
"gpt-4-1106-preview": {
"input_token_price_per_million": 10,
"output_token_price_per_million": 30
},
"gpt-4-turbo": {
"input_token_price_per_million": 10,
"output_token_price_per_million": 30
},
"gpt-4.1": {
"input_token_price_per_million": 2.0,
"output_token_price_per_million": 8.0
},
"gpt-4.1-mini": {
"input_token_price_per_million": 0.4,
"output_token_price_per_million": 1.6
"gemini-3-flash-preview": {
"input_token_price_per_million": 0.5,
"output_token_price_per_million": 3
},
"gpt-4.1-nano": {
"input_token_price_per_million": 0.1,
"output_token_price_per_million": 0.4
"o4-mini": {
"input_token_price_per_million": 1.1,
"output_token_price_per_million": 4.4
},
"gpt-4o": {
"input_token_price_per_million": 2.5,
"gpt-5.1": {
"input_token_price_per_million": 1.25,
"output_token_price_per_million": 10.0
},
"gpt-4o-mini": {
"input_token_price_per_million": 0.15,
"output_token_price_per_million": 0.6
},
"gpt-5-chat-latest": {
"gpt-5.1-None": {
"input_token_price_per_million": 1.25,
"output_token_price_per_million": 10.0
},
Expand All @@ -139,148 +27,108 @@
"input_token_price_per_million": 0.05,
"output_token_price_per_million": 0.4
},
"gpt-5.1": {
"input_token_price_per_million": 1.25,
"output_token_price_per_million": 10.0
},
"gpt-5.1-None": {
"input_token_price_per_million": 1.25,
"output_token_price_per_million": 10.0
},
"llama-3-1-405b-instruct": {
"input_token_price_per_million": 2.4,
"output_token_price_per_million": 2.4
},
"llama-3-1-8b-instruct": {
"input_token_price_per_million": 0.22,
"output_token_price_per_million": 0.22
},
"llama-3-2-1b-instruct": {
"input_token_price_per_million": 0.1,
"output_token_price_per_million": 0.1
},
"llama-3-2-3b-instruct": {
"input_token_price_per_million": 0.15,
"output_token_price_per_million": 0.15
},
"llama-3-3-70b-instruct": {
"input_token_price_per_million": 0.72,
"output_token_price_per_million": 0.72
},
"meta-llama/Llama-3-70b-chat-hf": {
"input_token_price_per_million": 0.88,
"output_token_price_per_million": 0.88
},
"meta-llama/Meta-Llama-3-70B-Instruct-Turbo": {
"input_token_price_per_million": 0.88,
"output_token_price_per_million": 0.88
},
"meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo": {
"input_token_price_per_million": 0.88,
"output_token_price_per_million": 0.88
},
"meta-llama/llama-3-8b-instruct": {
"input_token_price_per_million": 0.1,
"output_token_price_per_million": 0.1
},
"meta-llama_llama-3.1-405b-instruct": {
"input_token_price_per_million": 3.5,
"output_token_price_per_million": 3.5
},
"meta-llama_llama-3.3-70b-instruct": {
"input_token_price_per_million": 0.1,
"output_token_price_per_million": 0.32
},
"meta/codellama-34b-instruct": {
"input_token_price_per_million": 3.2,
"output_token_price_per_million": 3.2
"deepseek-reasoner": {
"input_token_price_per_million": 0.28,
"output_token_price_per_million": 0.42
},
"mistral-large-latest": {
"gemini-3.0-pro": {
"input_token_price_per_million": 2.0,
"output_token_price_per_million": 6.0
"output_token_price_per_million": 12.0
},
"mistral-medium": {
"input_token_price_per_million": 0.4,
"output_token_price_per_million": 2.0
"deepseek-v3.2": {
"input_token_price_per_million": 0.28,
"output_token_price_per_million": 0.42
},
"mistral-medium-latest": {
"input_token_price_per_million": 0.4,
"test": {
"input_token_price_per_million": 1.0,
"output_token_price_per_million": 2.0
},
"mistral-small-latest": {
"input_token_price_per_million": 0.1,
"output_token_price_per_million": 0.3
},
"mistralai/mistral-7b-instruct": {
"input_token_price_per_million": 0.2,
"output_token_price_per_million": 0.2
},
"mistralai/mixtral-8x7b-instruct": {
"input_token_price_per_million": 0.6,
"output_token_price_per_million": 0.6
},
"mistralai_devstral-2512:free": {
"input_token_price_per_million": 0.05,
"output_token_price_per_million": 0.22
},
"o4-mini": {
"input_token_price_per_million": 1.1,
"output_token_price_per_million": 4.4
"claude-sonnet-4-5": {
"input_token_price_per_million": 3.0,
"output_token_price_per_million": 15.0
},
"open-mistral-7b": {
"input_token_price_per_million": 0.15,
"output_token_price_per_million": 0.15
"claude-3-haiku-20240307": {
"input_token_price_per_million": 0.25,
"output_token_price_per_million": 1.25
},
"open-mistral-nemo": {
"gpt-4o-mini": {
"input_token_price_per_million": 0.15,
"output_token_price_per_million": 0.15
},
"open-mixtral-8x7b": {
"input_token_price_per_million": 0.6,
"output_token_price_per_million": 0.6
},
"openai_gpt-oss-120b": {
"input_token_price_per_million": 0.039,
"output_token_price_per_million": 0.19
},
"qwen_qwen3-235b-a22b-2507": {
"input_token_price_per_million": 0.071,
"output_token_price_per_million": 0.463
"glm-4.5-air": {
"input_token_price_per_million": 0.1,
"output_token_price_per_million": 0.3
},
"qwen_qwen3-coder": {
"input_token_price_per_million": 0.22,
"output_token_price_per_million": 0.95
"glm-4.6": {
"input_token_price_per_million": 0.28,
"output_token_price_per_million": 1.12
},
"qwen_qwen3-vl-235b-a22b-instruct": {
"input_token_price_per_million": 0.2,
"output_token_price_per_million": 1.2
"glm-4-air-250414": {
"input_token_price_per_million": 0.07,
"output_token_price_per_million": 0.07
},
"qwen_qwen3-vl-235b-a22b-thinking": {
"input_token_price_per_million": 0.3,
"output_token_price_per_million": 1.2
},
"qwen_qwen3-vl-32b-instruct": {
"input_token_price_per_million": 0.5,
"output_token_price_per_million": 1.5
},
"test": {
"input_token_price_per_million": 1.0,
"output_token_price_per_million": 2.0
},
"x-ai_grok-4.1-fast": {
"input_token_price_per_million": 0.2,
"output_token_price_per_million": 0.5
"gemini-2.5-flash-lite": {
"input_token_price_per_million": 0.1,
"output_token_price_per_million": 0.4
},
"x-ai_grok-code-fast-1": {
"input_token_price_per_million": 0.2,
"output_token_price_per_million": 1.5
"gemini-2.5-flash": {
"input_token_price_per_million": 0.3,
"output_token_price_per_million": 2.5
},
"xiaomi_mimo-v2-flash:free": {
"gemini-2.0-flash-001": {
"input_token_price_per_million": 0.1,
"output_token_price_per_million": 0.3
"output_token_price_per_million": 0.4
},
"z-ai_glm-4.7": {
"input_token_price_per_million": 0.4,
"qwen_qwen3-vl-32b-instruct": {
"input_token_price_per_million": 0.5,
"output_token_price_per_million": 1.5
}
},
"z-ai_glm-4.7": {
"input_token_price_per_million": 0.4,
"output_token_price_per_million": 1.5
},
"qwen_qwen3-vl-235b-a22b-instruct": {
"input_token_price_per_million": 0.2,
"output_token_price_per_million": 1.2
},
"qwen_qwen3-coder": {
"input_token_price_per_million": 0.22,
"output_token_price_per_million": 0.95
},
"x-ai_grok-code-fast-1": {
"input_token_price_per_million": 0.2,
"output_token_price_per_million": 1.5
},
"xiaomi_mimo-v2-flash:free": {
"input_token_price_per_million": 0.1,
"output_token_price_per_million": 0.3
},
"openai_gpt-oss-120b": {
"input_token_price_per_million": 0.039,
"output_token_price_per_million": 0.19
},
"qwen_qwen3-235b-a22b-2507": {
"input_token_price_per_million": 0.071,
"output_token_price_per_million": 0.463
},
"x-ai_grok-4.1-fast": {
"input_token_price_per_million": 0.2,
"output_token_price_per_million": 0.5
},
"mistralai_devstral-2512:free": {
"input_token_price_per_million": 0.05,
"output_token_price_per_million": 0.22
},
"meta-llama_llama-3.3-70b-instruct": {
"input_token_price_per_million": 0.1,
"output_token_price_per_million": 0.32
},
"meta-llama_llama-3.1-405b-instruct": {
"input_token_price_per_million": 3.5,
"output_token_price_per_million": 3.5
}
}
Loading