[jamesrochabrun]

Summary

Adds support for the models and service tier announced at OpenAI DevDay 2026 (September 29, 2026).

• New models in Model:
  • .gpt6Astra → gpt-6-astra
  • .gpt61Sol → gpt-6.1-sol (1,050,000 token context window, 128,000 max output tokens)
  • .gpt56Sol → gpt-5.6-sol (Ultrafast preview model)
• New ServiceTier enum (auto, default, flex, priority, ultrafast) so the Ultrafast tier is discoverable. The existing serviceTier: String? parameter on ModelResponseParameter is unchanged, so this is source-compatible — pass ServiceTier.ultrafast.rawValue.
• README: documents the GPT-6 models and adds an Ultrafast usage example.

GPT-6 Luna is intentionally not included: OpenAI has not published its API model ID — it is currently only reachable through the Decisions API, which is in limited preview without public documentation. Agents API and Decisions API support will follow in separate PRs once implemented/documented.

Usage

let parameters = ModelResponseParameter(
    input: .text("Summarize this ticket in one sentence."),
    model: .gpt6Astra,
    serviceTier: ServiceTier.ultrafast.rawValue
)

let response = try await service.responseCreate(parameters)

Testing

• swift build succeeds
• All 101 existing tests pass
• swiftformat (repo rules) reports no violations

🤖 Generated with Claude Code