Files
cc-switch/tests/config
Jason 50a9d103d3 chore(presets): refresh preset models to the latest vendor lineups
Move first-party, sponsor and relay presets to current model IDs so a
pasted Key works without editing the model list.

- Claude: opus/sonnet 5 -> 5.5, haiku 4.5 -> 5.5, fable 5 -> 5.1
- GPT: gpt-5.6-sol / gpt-6-sol -> gpt-6.1-sol; add gpt-6-astra/luna
- Gemini 3.6 Flash -> 3.8 Flash, 3.1 Pro preview -> 4 Argon;
  Grok 4.5 -> 4.7; GLM 5.1/5.2 -> 5.3 (+5.3 Flash);
  DeepSeek V4 Flash -> V4.1 Flash; Doubao 2.1 Pro 260628 -> 260915;
  StepFun Step Plan -> step-5-preview; add Kimi K3.1 Pro, Qwen3.8 Max,
  MiniMax M3.1 Flash where the site serves them
- Drop retired models: NVIDIA moonshotai/kimi-k3 (EOL 2026-10-15,
  now kimi-k2.6 without unverified Codex reasoning levels),
  MiMo V2.5 / V2.5 Pro, Claude Sonnet 4.5, MiniMax M2.7 on Tencent
  Token Plan Enterprise Pro
- Relay presets follow each site's own ID spelling and catalog;
  DIT.ai, BuyToken and hao.ai are left as configured from their docs;
  Soshow, SiliconFlow CN, OpenCode Zen and the Ark rolling alias stay
- MoArk Codex keeps the tested Kimi-K2.7-Code; K3.1 Pro and Qwen3.8
  Max stay out of its Codex catalog until their levels are verified

Tests updated to the new IDs; cross-app invariants unchanged.
2026-10-24 22:56:15 +08:00
..