Problem#
Two OpenAI chat rows have capabilities.reasoning: true but reasoning: null, with no factSources.reasoning entry:
gpt-6.2-astra
gpt-6.2-sol
Every other reasoning row of openai, anthropic, gemini, grok, groq, mistral and byteplus has either a reasoning object or a factSources.reasoning entry (source-silent or docs-derived). Snapshot 2026-11-21.
Impact#
TanStack AI types the reasoning options of a new model from reasoning.mode, mandatory and efforts. For these two it has nothing to read.
Ask#
Fill reasoning for both rows from the OpenAI model pages, or add the factSources.reasoning entry.
Acceptance#
GET /v1/models/openai/gpt-6.2-astra and /gpt-6.2-sol return a reasoning object or a factSources.reasoning entry.