fix: [missing-model] ofox: openai/gpt-6-astra

[github-actions]

Summary

Adds the missing Ofox catalog entry for openai/gpt-6-astra, using existing lab metadata via base_model and Ofox-verified flat pricing.

Changes

• New providers/ofox/models/openai/gpt-6-astra.toml
  • base_model = "openai/gpt-6-astra" (lab file already present)
  • reasoning_options: effort low / medium / high / xhigh / max (matches OpenAI first-party and peer relays; no none)
  • cost: input 10, output 50, cache_read 1, cache_write 12.5 (USD per 1M tokens)
  • No [[cost.tiers]] — Ofox catalog/model page publish a single flat rate band
  • [provider]: npm = "@ai-sdk/openai", api = "https://api.ofox.ai/v1" (same pattern as other Ofox OpenAI models)

Evidence

Claim	Source	What it establishes
Model exists on Ofox as openai/gpt-6-astra	ofox.ai/models/openai/gpt-6-astra	Listed chat model with OpenAI-compatible /v1 usage
Pricing $10 / $50 / $1 cache read / $12.5 cache write	Same page + Ofox catalog API (pricing.input=0.00001, etc.)	Flat Ofox rates; no long-context tier fields in catalog
Effort levels low–max	OpenAI GPT-6 Astra docs	reasoning.effort supports low, medium, high, xhigh, max
Lab metadata / limits	Existing models/openai/gpt-6-astra.toml	Inherited name, modalities, 1.05M context, 128K output, reasoning=true

Validation

• Cross-checked Ofox model page and live catalog API pricing against OpenAI docs for effort values.
• Followed existing Ofox OpenAI TOML patterns (gpt-5.6-terra, gpt-5.5, gpt-5.4-pro note on no tiers when catalog is flat).
• Did not run bun validate (no shell per instructions).

Review notes

• Ofox does not list a gpt-6-astra-fast variant; only the base model was added.
• OpenAI first-party documents a >272K input surcharge; Ofox’s public catalog does not, so tiers were omitted intentionally (same rationale as Ofox gpt-5.4-pro).

Closes #6337

Automated by the issue fixer: https://github.com/anomalyco/models.dev/actions/runs/33947120290