Qwen3.7 Flash is a vision-language reasoning model from Alibaba. It is suited for multimodal agents, visual coding, search, and computer interaction, with strengths in object recognition, spatial understanding, and real-world...
Tested on 5 benchmarks · BenchGecko score 44.3. Top scores: OTIS Mock AIME 2024-2025 (86.7%), GPQA diamond (76.4%), FrontierMath-Tiers-1-3-v2-Private (19.3%).
Mock AIME (American Invitational Mathematics Exam) problems from OTIS. Tests mathematical competition performance.
Graduate-level science questions written by PhD experts. Diamond subset contains questions where experts disagree, testing deep understanding.
Tactical chess puzzles testing pattern recognition and multi-move calculation. Measures strategic reasoning ability.
- Typemultimodal
- Context1.0M tokens (~500 books)
- ReleasedJul 2026
- LicenseProprietary
- StatusActive
- Cost / Message~$0.000
Frequently Asked Questions
Related Models
Claude Sonnet 4 · AnthropicGemini 2.0 Flash · Google DeepMindGemini 2.0 Flash (Dec 2024) · Google DeepMindNemotron 3 Ultra · NVIDIAGrok 3 · xAIKey facts · as of 2026-10-15
- Qwen3.7 Flash by Alibaba Qwen. BenchGecko score 44.3, rank 180 of 319 scored models (normalized average of public benchmark scores).
- List price $0.0300 input · $0.13 output per 1M tokens (as of 2026-10-15).
- Sold by 2 providers (as of 2026-10-15): Alibaba $0.0300 in / $0.13 out · Alibaba $0.0345 in / $0.15 out. Every provider
How to cite · data as of 2026-10-15
Qwen3.7 Flash · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-15. https://benchgecko.ai/model/qwen3-7-flash
Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP