GPT-5.6 models missing max reasoning effort variant

[javargasm]

Description

OpenAI's reasoning_options for GPT-5.6 family models (gpt-5.6, gpt-5.6-terra, gpt-5.6-luna, gpt-5.6-sol) include max as a valid reasoning_effort value, but OpenCode only exposes efforts up to xhigh.

The versionedGpt5ReasoningEfforts function in transform.ts returns OPENAI_GPT5_2_PLUS_EFFORTS for all versions >= 2, which caps at xhigh. GPT-5.6+ models need an additional max tier.

Expected behavior

When selecting a GPT-5.6 family model, the reasoning effort selector should include: none, low, medium, high, xhigh, max.

Actual behavior

The selector only shows up to xhigh. The max effort variant is missing.

Steps to reproduce

1. Configure OpenAI provider with a GPT-5.6 model (e.g. gpt-5.6-terra)
2. Open the reasoning effort selector
3. Observe max is not available

Additional context

OpenAI API reference for gpt-5.6-terra shows:

"reasoning_options": [{
  "type": "effort",
  "values": ["none", "low", "medium", "high", "xhigh", "max"]
}]

[github-actions]

This issue might be a duplicate of or related to existing issues. Please check:

• fix(copilot): claude-sonnet-4.6 max reasoning effort not available; model key mismatch on refresh #33607: Same root cause pattern — max reasoning effort missing for a specific model family, traced to transform.ts effort filtering logic
• GLM-5.2 on OpenAI-compatible providers: TUI variant switcher shows max (invalid reasoning_effort) instead of xhigh #34278: Related — TUI variant switcher and max/xhigh effort availability mismatch on OpenAI-compatible providers

If your issue is specifically about GPT-5.6 models and the max tier not appearing in the effort selector, it may be worth coordinating with or referencing those issues.

[rekram1-node]

this is the correct fix:
#35985

[SGSStore7]

+1 and ultra is missed too

[elijahmorg]

Temporary workaround until the proper fix is released.

opencode.json

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "openai": {
      "models": {
        "gpt-5.6-luna": {
          "variants": {
            "max": {
              "reasoningEffort": "max"
            }
          }
        },
        "gpt-5.6-terra": {
          "variants": {
            "max": {
              "reasoningEffort": "max"
            }
          }
        },
        "gpt-5.6-sol": {
          "variants": {
            "max": {
              "reasoningEffort": "max"
            }
          }
        }
      }
    }
  }
}

Luna is not working for me in opencode regardless of this change. I am unsure if that is my corporate codex account or not...

[ririnto]

ultra is not effort value. it's just Orchestrator mode with max effort.