Antigravity model quota grouping/mapping appears incorrect

[Pixtella]

Antigravity’s current model list appears to be:

• Gemini 3.1 Pro (High)
• Gemini 3.1 Pro (Low)
• Gemini 3 Flash
• Claude Sonnet 4.6 (Thinking)
• Claude Opus 4.6 (Thinking)
• GPT-OSS 120B (Medium)

Based on observation, quota groups seem to be:

• Pro High + Pro Low share one quota window
• Gemini Flash has its own quota window
• Claude Sonnet + Claude Opus + GPT-OSS share one quota window

CodexBar currently renders only two bars for Antigravity, and these seem to be mapped incorrectly (e.g. the “Claude” bar actually shows quota for Gemini Pro).

Likely cause: model selection/parsing heuristics no longer match current Antigravity model labels/groupings (especially with “(Thinking)” variants), so bar-to-model mapping is wrong.

Suggested fix:

• Update parsing/grouping logic to reflect current Antigravity model families and shared quotas.
• Avoid assigning non-Claude windows to the “Claude” bar when matching is ambiguous.

Likely files to edit

• Sources/CodexBarCore/Providers/Antigravity/AntigravityStatusProbe.swift

  • model label matching / selection logic
  • quota grouping + fallback behavior

• Sources/CodexBarCore/Providers/Antigravity/AntigravityProviderDescriptor.swift

  • UI slot labels (sessionLabel, weeklyLabel, opusLabel) and expected mapping assumptions

• docs/antigravity.md

  • update documented model mapping + fallback order to match current Antigravity model set

[Pixtella]

CodexBar version: 0.18.0
macOS: 26.3.1 (Tahoe), Apple Silicon (arm64)

[skainguyen1412]

@Pixtella You can try my fix #562