v2.1.23

✨ 本次更新

• 新增对 GPT-6、GPT-6.1 及 Gemini 2.5 Pro 模型支持，并将默认模型升级为 gpt-6.1-sol。
• 模型列表现在会自动刷新并与官方目录实时同步，解决了模型列表可能出现过时或不准确的问题。
• 为 Gemini 和 Anthropic 模型引入了 Token 计数和“思考”（Thinking）过程支持，功能体验更接近原生。
• 优化了响应处理逻辑，修复了部分情况下响应被意外截断或元数据丢失的问题，并提升了流式传输的稳定性。
• 改进了与各类客户端的兼容性，支持为特定客户端版本提供完整的 Codex 模型目录元数据。

What's New

• Added support for GPT-6, GPT-6.1, and Gemini 2.5 Pro, with gpt-6.1-sol now set as the default model.
• The model catalog now refreshes automatically and stays in sync with the official backend, ensuring you always have the most up-to-date model list without stale data.
• Introduced token counting and "thinking" (reasoning) parity for Gemini and Anthropic models for a more native experience.
• Improved response handling by fixing issues with unexpected truncation and missing metadata, while also enhancing the stability of streaming responses.
• Enhanced compatibility for various coding clients by serving the full Codex model catalog and metadata for client version requests.

完整提交记录 / Full commit list

• feat(config): upgrade default model to gpt-6.1-sol (#895)
• fix(responses): passthrough store, metadata and truncation (#893)
• feat(gemini,anthropic): add countTokens and thinking parity (#892)
• feat(antigravity): add gemini-2.5-pro to supported models (#882)
• fix(responses): normalize string input and align stream defaults (#866) (#883)
• feat(models): serve Codex catalog shape for client_version requests (#891)
• feat(models): refresh official catalog automatically (#890)
• fix(gemini): serve live model catalog on /v1beta/models (#888)
• feat(models): add gpt-6.1 and gpt-6 catalog entries (#884)
• feat(models): sync official Codex catalog snapshot (#887)
• fix(models): serve live backend catalog without stale static fallback (#885)