Qwen 3.5 Plus
The Qwen3.5 native vision-language series Plus models are built on a hybrid architecture that integrates linear attention mechanisms with sparse mixture-of-experts models, achieving higher inference efficiency.
text + image + file → text · made by Alibaba
Sold by 4 ways
| Seller | Lane | Rate |
|---|---|---|
| AtlasCloud aggregator | standard | $0.4per Mtok in$2.4per Mtok out$0.04per Mtok cachedapi.atlascloud.ai · read 2026-09-20 |
| Vercel AI Gateway aggregator | standard | $0.4per Mtok in$2.4per Mtok out$0.04per Mtok cached$0.5per Mtok cache writeai-gateway.vercel.sh · read 2026-08-25 |
| ElectronHub aggregator | standard | $0.8per Mtok in$3per Mtok outapi.electronhub.ai · read 2026-10-31 |
| PPIO 0 ways | $4per Mtok in$24per Mtok out | |
| PPIO aggregatornot on the seller's list today | standard | $4per Mtok in$24per Mtok outapi.ppinfra.com · read 2026-10-08 |
Measured 14 standings
| Place | Board | Metric | Score |
|---|---|---|---|
| 9thof 23 | Cl bench — Epoch AI | Overall | 0.198 |
| 11thof 17 | Cl bench life — Epoch AI | Overall | 0.124 |
| 38thof 101 | FrontierMath (Tiers 1-3) | mean_score | 21.034 |
| 50thof 72 | Frontiermath tier 4 — Epoch AI | mean_score | 2.083 |
| 66thof 71 | Vending bench 2 — Epoch AI | Score | 54.4 |
| 74thof 93 | Simpleqa verified — Epoch AI | mean_score | 25.351 |
| 80thof 142 | Mystery game puzzles — Epoch AI | mean_score (none) | 17 |
| 87thof 135 | Webdev arena — Epoch AI | Arena Score | 1398.92 |
| 92ndof 306 | OTIS Mock AIME 2024-2025 | mean_score | 86.667 |
| 100thof 132 | Ale bench — Epoch AI | Performance | 621.92 |
| 101stof 326 | GPQA Diamond | mean_score | 84.848 |
| 124thof 197 | Lmca — Epoch AI | Score | 36.36 |
| 130thof 154 | Chess puzzles — Epoch AI | mean_score | 8 |
| 132ndof 235 | Dtbench — Epoch AI | Accuracy | 80.53 |
About
Qwen 3.5 Plus — a text model from Alibaba, sold by 3 companies from $0.4 in and $2.4 out per million tokens, placed 92nd of 306 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 February 2026, trained on material up to April 2025. Its sellers say it can reason step by step and call a tool. The catalogue files it under extract, evaluate, reasoning and agents. Three companies sell it. The cheapest is $0.4 in and $2.4 out per million tokens at AtlasCloud. It has been measured on 14 boards, and stands best at 92nd of 306 on OTIS Mock AIME 2024-2025.
Every current figure
- Maker
- Alibaba
- Register
- model
- Takes
- text + image + file
- Returns
- text
- Context
- 1,000,000 tokens
- Longest answer
- 250,000 tokens
- Published
- February 2026
- Knowledge to
- April 2025
- Licence
- not read
- Sellers
- 3
- Maker's own price
- not read
- Price
- $0.4 in and $2.4 out per million tokens — AtlasCloud
- Boards
- 14
- Best place
- 92nd of 306 — OTIS Mock AIME 2024-2025
Known as 2 names
qwen/qwen3.5-plusalibaba/qwen3.5-plus