- Streaming/Non-Streaming
/v1/chat/completions,/v1/completions,/v1/messages- configurable TTFT/per chunk latency via request header or request body
- serverd mode name mocking
A schema v2 model catalog for swiss, served by GitHub Pages at https://modelsphere.github.io/mock-llmd/index-v2.json.
catalog.yaml: the catalog'ssiteandrepomodels/<name>/metadata.yamland one<id>.yamlper variant, as in model-catalog
go run ./hack/build-site/main.go writes site/ (needs the full git history).