Qwen3.5-Flash
The Qwen3.5 native vision-language Flash models are built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency.
text + image + file → text · made by Alibaba
Sold by 3 ways
| Seller | Lane | Rate |
|---|---|---|
| Nous Research aggregator | standard | $0.065per Mtok in$0.26per Mtok outinference-api.nousresearch.com · read 2026-09-12 |
| OpenRouter aggregator | standard | $0.065per Mtok in$0.26per Mtok outopenrouter.ai · read 2026-08-24 |
| Vercel AI Gateway aggregator | standard | $0.1per Mtok in$0.4per Mtok out$0.01per Mtok cached$0.12per Mtok cache writeai-gateway.vercel.sh · read 2026-08-25 |
Measured 17 standings
| Place | Board | Metric | Score |
|---|---|---|---|
| 50thof 127 | Mystery game puzzles — Epoch AI | mean_score | 20 |
| 50thof 60 | Vending bench 2 — Epoch AI | Score | 462.685 |
| 52ndof 161 | Dtbench — Epoch AI | Accuracy | 82.93 |
| 60thof 105 | Vectara Hallucination Leaderboard | Hallucination Rate | 10.5 |
| 62ndof 72 | Frontiermath tier 4 — Epoch AI | mean_score | 0 |
| 64thof 101 | FrontierMath (Tiers 1-3) | mean_score | 6.207 |
| 66thof 80 | Simpleqa verified — Epoch AI | mean_score | 20.32 |
| 68thof 123 | Lmca — Epoch AI | Score | 29.14 |
| 78thof 222 | Chess puzzles — Epoch AI | mean_score | 21 |
| 83rdof 106 | Frontiermath tiers 1 3 v2 — Epoch AI | mean_score (none) | 18.246 |
| 89thof 290 | OTIS Mock AIME 2024-2025 | mean_score | 84.444 |
| 108thof 112 | Ale bench — Epoch AI | Performance | 221.8 |
| 110thof 313 | GPQA Diamond | mean_score | 82.323 |
| 114thof 121 | Webdev arena — Epoch AI | Arena Score | 1238.12 |
| 118thof 128 | LMArena · WebDev | Score (Elo) | 1238 |
| 152ndof 402 | LMArena · Text | Score (Elo) | 1397 |
| 352ndof 656 | Epoch capabilities index — Epoch AI | ECI Score (none) | 144.03 |
About
Qwen3.5-Flash — a text model from Alibaba, sold by 3 companies from $0.065 in and $0.26 out per million tokens, placed 89th of 290 on OTIS Mock AIME 2024-2025.
It takes text, images and files and returns text, with a context window of 1,000,000 tokens. It was published in January 2026. Its sellers say it can reason step by step and call a tool. The catalogue files it under extract, reasoning, code and agents. Three companies sell it. The cheapest is $0.065 in and $0.26 out per million tokens at Nous Research. It has been measured on 17 boards, and stands best at 89th of 290 on OTIS Mock AIME 2024-2025.
Every current figure
- Maker
- Alibaba
- Register
- model
- Takes
- text + image + file
- Returns
- text
- Context
- 1,000,000 tokens
- Published
- January 2026
- Licence
- not read
- Sellers
- 3
- Maker's own price
- not read
- Price
- $0.065 in and $0.26 out per million tokens — Nous Research
- Boards
- 17
- Best place
- 89th of 290 — OTIS Mock AIME 2024-2025
Known as 3 names
Qwen 3.5 Flashqwen/qwen3.5-flash-02-23alibaba/qwen3.5-flash