mirror of
https://github.com/diegosouzapw/OmniRoute.git
synced 2026-08-02 21:32:10 +03:00
4038e4de4a
* fix(pricing+test): price gpt-5.x-pro openai models + align opencode-go discovery test
Two pre-existing reds on the release's full unit suite (only surface under __RUN_ALL__):
1. catalog-updates-v3x: the provider sweep added gpt-5.5-pro / gpt-5.4-pro to the
openai registry but no pricing rows, so they resolved to $0 and tripped the
'Every OpenAI registry model resolves a non-zero pricing row' gate. Add both
under the openai pricing block (mirroring their base family tier until OpenAI
publishes a distinct pro rate).
2. provider-models-route: opencode-go discovery now (a) stamps owned_by on each
discovered model (fallback = provider id) and (b) probes ${base}/v1/models then
${base}/models (T39 multi-endpoint). The test fixtures predated both — add
owned_by to the expected model and bump the fail-path fetchCalls 1 -> 2. Test-only
alignment to the intentional route behavior.
* chore(quality): rebaseline file-size for #4355 (pricing +11, route test +2)