vava-nessa/free-coding-models v0.4.3

on GitHub

Added

• [nvidia] stepfun-ai/step-3.7-flash (Step 3.7 Flash, S+, 256k ctx) — new model appeared on NVIDIA NIM free tier on May 29, 2026
• [qwen] qwen3.7-max (Qwen3.7 Max, S+, 80.0% SWE, 1M ctx) — new flagship model confirmed on OpenRouter API (created 2026-05-20)
• [openrouter] moonshotai/kimi-k2.6:free (Kimi K2.6, S+, 262k ctx) — newly available with :free tier on OpenRouter

Removed

• [cloudflare] @cf/qwen/qwq-32b (QwQ 32B) — upstream deprecation: Groq removed qwen-qwq-32b in July 2025, Cloudflare has mirrored this removal
• [openrouter] minimax/minimax-m2.5:free (MiniMax M2.5) — free tier removed from OpenRouter
• [openrouter] deepseek/deepseek-v4-flash:free (DeepSeek V4 Flash) — free tier removed from OpenRouter
• [openrouter] baidu/cobuddy:free (Baidu CoBuddy) — free tier removed from OpenRouter

Fixed

• [github-models] meta/llama-4-scout-17b-16e-instruct ctx 10M → 1M — GitHub Models caps context at 1M, not 10M

Changed

• Model count: 155 → 153 (net -2)
  • nvidia: 26 → 27 (+1)
  • cloudflare: 14 → 13 (-1)
  • openrouter: 26 → 24 (-3+1)
  • qwen: 11 → 12 (+1)
• All other providers verified current (no changes needed)

Notes

• gemini-2.5-pro/flash/flash-lite still active until October 16, 2026 — schedule removal for next audit cycle
• Mistral deprecations in July/August 2026: devstral-2512, magistral-medium-2509 — schedule removal for next audit cycle
• SambaNova: gemma-3-12b-it is now PREVIEW (not production) — still usable but may be removed at short notice