Qwen 3.7
#6 in Open-Source-Spraakmodellealibaba · v3.7 · siet 2026-05-19 (Qwen3.7-Max); 2026-06-03 (Qwen3.7-Plus) · 60× · tolest 29. Juni 2026
68
Momentum
Qwen 3.7 is Alibaba's proprietary model series within the Qwen family, comprising two variants: Qwen3.7-Max (text-only flagship coding agent) and Qwen3.7-Plus (multimodal with image and video understanding). Both models are closed-weight and available exclusively via API through Alibaba Cloud Model Studio – no download, no self-hosting. The focus is on agent-centric, long-horizon autonomous task execution across hundreds to thousands of steps.
Momentum-Verloop
04.04.03.07.
Features
| Key Benchmark (%) | GPQA Diamond: 92.4 (vs. Opus-4.6 Max: 91.3); SWE-Verified: 80.4 |
| Context Window (Tokens) | 1,000,000 tokens (max. output 65,536 tokens) |
| License | Proprietary, closed-weight (no open-weight release, no Apache 2.0) |
| Multimodality | Qwen3.7-Max: text-only; Qwen3.7-Plus: multimodal (text, image, video), unifying vision & language as an agent foundation |
| Platform | Alibaba Cloud Model Studio (API), Qwen Chat; also OpenRouter, Together AI, Fireworks |
| Price per 1M Tokens | Qwen3.7-Max: $2.50 input / $7.50 output (list price); temporary promo $1.25/$3.75 |
| Release Date | Qwen3.7-Max: API live May 19, 2026, formal announcement May 20, 2026 (Alibaba Cloud Summit Hangzhou) |