Reasoning effort was a fixed per-model setting stored at provider-specific paths in chat_model_configs.options (e.g. provider_options.openai.reasoning_effort), so offering the same model at multiple efforts required duplicate model configs. Each model config now carries a top-level reasoning_effort {default, max} pair instead.

This PR is behavior-preserving: generation resolves the configured default (clamped to max and the provider's supported set), which matches the old fixed value after migration. Per-turn user selection lands in the follow-up PR.

Changes

• codersdk: new ChatModelReasoningEffortConfig (default/max) on ChatModelCallConfig; the per-provider effort fields (openai.reasoning_effort, anthropic.effort, openaicompat.reasoning_effort, openrouter/vercel reasoning.effort) are removed.
• Validation: values must be in the provider's runtime-supported set (openai/openaicompat minimal..xhigh, anthropic low..max, openrouter low..high, vercel none..xhigh) with default <= max on the global scale none < minimal < low < medium < high < xhigh < max; a single value mirrors into the other. Updates fail when the config's provider has been soft-deleted.
• Migration (000537): rewrites chat_model_configs.options so the legacy effort value (selected provider-aware via ai_providers.type, with a COALESCE fallback) becomes both default and max, stripping the legacy keys. Also adds chats.last_reasoning_effort, chat_messages.reasoning_effort, and chat_queued_messages.reasoning_effort, which stay unused until the per-turn follow-up. Down migration restores the provider-appropriate legacy path.
• chatd: new chatprovider.ResolveReasoningEffort clamps and snaps the effective effort into the provider's supported set before injecting it into the fantasy provider options (azure routes via openai, bedrock via anthropic).
• site: compile fallout only; the admin form schema loses the old per-provider effort selects, known-model catalog defaults populate the new pair, and form logic validates default <= max. New admin UI placement and the chat slider land in follow-ups.

Stack: this PR → per-turn API (#26975) → models page UI (#26976, on this PR) and chat slider (#26977, on #26975).

🤖 Generated by Coder Agents on behalf of @DanielleMaywood