meadow-kun/llm-speed bench
Run llm-speed bench in CI, upload a signed run, and post the result as a PR comment.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| model | Model identifier passed to the chosen backend (e.g. mlx-community/Qwen2.5-7B-Instruct-4bit, qwen/qwen3-coder-next). | yes | — |
| backend | One of: ollama, llama.cpp, vllm, mlx, exllamav2, hosted-api. | yes | — |
| api-base | Base URL for the llm-speed API. | no | https://api.llm-speed.com |
| api-key | Optional submitter API key. Leave empty for anonymous uploads. | no | "" |
| anon | Submit anonymously (no signing key). Default false. | no | false |
| workload | Workload preset to run. | no | quick |
| comment-on-pr | Post the result back as a PR comment when running on a pull_request event. | no | true |
Outputs
| name | description |
|---|---|
| run-id | The r_xxxx slug returned by the API. |
| decode-tps | Decode tokens-per-second (numeric). |
| run-url | Full https URL to the run page on llm-speed.com. |
| badge-url | Full https URL to the badge SVG. |