Greptile Summary

This automated PR syncs model provider data from the upstream PublicProviderConf source, updating providers.json with new models and revised metadata for existing ones.

• New models added: claude-opus-4-8 (Anthropic), kat-dev-72b-exp (Kwaipilot), and three Xiaomi MiMo-V2.5-TTS variants; the Anthropic addition introduces a new experimental.modes.fast cost tier pattern.
• Metadata updates: 36 existing models received field corrections including context-window expansions for DeepSeek models, minor price adjustments for several OpenAI models (e.g., gpt-4o-mini, o4-mini), and gpt-3.5-turbo's cache_read cost changed from 1.25 to 0.
• No providers removed or added: the same 16 providers are present; the apparent churn in the diff is due to provider ordering changing (deepseek moved to first position).

Confidence Score: 4/5

Safe to merge; all changes are to a static JSON data file with no structural issues.

The JSON is valid and no providers or models were accidentally dropped. Two data quality observations are worth noting: the new kat-dev-72b-exp model is missing standard fields like cost and modalities (though the frontend handles this gracefully), and gpt-3.5-turbo's cache_read cost jumped to zero — an unusual change that could reflect an upstream data error rather than a real pricing update.

frontend/src/features/models/data/providers.json — the kwaipilot/kat-dev-72b-exp entry and the gpt-3.5-turbo cache_read value are worth a quick manual check.

Important Files Changed

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[PublicProviderConf upstream source] -->|weekly sync workflow| B[providers.json]
    B --> C{Changes in this PR}
    C --> D[New models added\nclaudeOpus48 · katDev72bExp\nmimo-v2.5-tts x3]
    C --> E[Metadata updated\n36 models updated\ncosts · context windows · flags]
    C --> F[Provider reordering\ndeepseek moved to 1st]
    D --> G1[claude-opus-4-8\nnew experimental fast-mode tier]
    D --> G2[kat-dev-72b-exp\nmissing cost/modalities fields]
    E --> H[gpt-3.5-turbo\ncache_read 1.25 → 0]

_{Reviews (1): Last reviewed commit: "chore: sync model developers data" | Re-trigger Greptile}