[c121914yu]

Summary

Extend the existing general-purpose provider presets and keep the list free of specialist OpenAI models, dated Qwen snapshots, and Qwen US-region duplicates.

• Add Ant Ling Ling-3.1-flash, Doubao doubao-seed-2-1-pro-260915 / doubao-seed-2-1-lite-260915, MiniMax MiniMax-M3.1-Flash, Gemini gemini-4-argon, Baidu ernie-4.5-vl-28b-a3b, and Spark spark-x2.5 / spark-x2.5-1.7b / spark-x2.5-4b.
• Remove Qwen presets with date or snapshot suffixes, including -YYYY-MM-DD and -MMDD, and exclude -us variants. Keep the corresponding family IDs, such as qwen3.8-max and qwen3.7-plus, with their existing configurations.
• Add a regression check for the Qwen naming rule and remove obsolete snapshot-order assertions. Update both copies of the model-provider skill so later refreshes skip these variants.
• Exclude OpenAI gpt-5.6-cyber (purpose-built cybersecurity) and gpt-rosalind-research (life-sciences research) from this general-purpose preset refresh. Both require separately approved access. Update both model-provider skill copies to skip these specialist families even when they are generally available. Official references: GPT-5.6 Cyber and Rosalind pricing and access.

Validation

• bun run test: 57 files / 406 tests passed.
• bun tsc --noEmit: passed.
• Model inventory and git diff --check: passed.
• Confirmed every removed Qwen variant has a retained model without the suffix, and all 51 retained Qwen configurations are unchanged.
• Both model-provider skill copies passed quick_validate.py.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[github-actions]

✅ Build and publish successful - Preview Server image:

ghcr.io/labring/fastgpt-plugin-pr:server_ee24f843c357f6fc1816aef720c35c73b29a6930

🕒 Time: 2026-10-24 22:56:40 (UTC+8)