AINews Portal

Best AI Reasoning Models

8 models — live arena Elo, API pricing, downloads, and our news-mention tracking. Synced daily. JSON is available at /api/models.

#ModelUseArena EloIn / Out $/MContextHF DLs7d NewsReleased
1Qwen3 235B A22B Thinking 2507qwen🧮 Reasoning1416$0.45 / $3.50262K7.2K—16mo
2Qwen3 VL 235B A22B Thinkingqwen🧮 Reasoning1401$0.40 / $4.00131K81.5K—14mo
3Qwen3 Next 80B A3B Thinkingqwen🧮 Reasoning1368$0.15 / $1.20262K45.4K—14mo
4Qwen Plus 0728 (thinking)qwen🧮 Reasoning—$0.26 / $0.781.0M——14mo
5Qwen3 Max Thinkingqwen🧮 Reasoning—$0.78 / $3.90262K——9mo
6Qwen3 30B A3B Thinking 2507qwen🧮 Reasoning—$0.20 / $2.40131K91.4K—14mo
7Qwen3 VL 30B A3B Thinkingqwen🧮 Reasoning—$0.29 / $1.00131K7.2K—13mo
8Qwen3 VL 8B Thinkingqwen🧮 Reasoning—$0.18 / $2.10256K134.9K—13mo

Elo: LMArena per-category arenas (CC-BY) — image/video/search models are ranked by their own arena, not text. Pricing: OpenRouter/LiteLLM. Use-case labels are auto-classified — suggest fixes on any model page. Methodology.