Lightweight multimodal Qwen model for high-throughput text, image, and video tasks
Specification
alibaba-cn listing
Context
1M
Max output
66K
Released
2026-07-15
Knowledge cutoff
—
Retires
—
Open weights
no
Input
text, image, video
Output
text
Price
US dollars per million tokens · alibaba-cn list price
Input
$0.0296
Output
$0.1185
Cache read
$0.003
Cache write
$0.037
Long-context price tiers
the rate above each context threshold
| Above | Price | $/M |
|---|---|---|
| 256K | Cache read | $0.04 |
| 256K | Cache write | $0.25 |
| 256K | Input | $0.20 |
| 256K | Output | $0.80 |
Quality
5 benchmarks · 1 source
Composite
64th
percentile of 334 scored models
Rank
120 / 334
±0.08 sd
Evidence
5 × 1
one source, or fewer than 3 benchmarks
Effort range
—
one configuration only
reasoning72nd3/4 bench
math56th2/7 bench
Every score
one row per source and configuration — nothing averaged away
| Benchmark | Score | Configuration | Source | Run |
|---|---|---|---|---|
| OTIS Mock AIME 2024-2025 | 86.7 ±5.1 | default | Epoch AI | 2026-08-07 |
| GPQA diamond | 82.3 ±2.7 | default | Epoch AI | 2026-08-07 |
| Chess Puzzles | 23.0 ±4.2 | default | Epoch AI | 2026-08-07 |
| Mystery Game Puzzles | 15.0 ±3.6 | default | Epoch AI | 2026-08-28 |
| FrontierMath-Tiers-1-3-v2 | 19.3 ±2.3 | default | Epoch AI | 2026-08-29 |
The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.
Available from 13 hosts
| Host | In $/M | Out $/M | Cache rd | Cache wr | Context | Output | Retires |
|---|---|---|---|---|---|---|---|
| alibaba-cnvendor | $0.0296 | $0.1185 | $0.003 | $0.037 | 1M | 66K | — |
| aihubmix | $0.0282 | $0.1128 | $0.0056 | $0.0352 | 991K | 64K | — |
| empiriolabs | $0.03 | $0.13 | $0.006 | — | 1M | 66K | — |
| kilo | $0.03 | $0.13 | $0.006 | $0.038 | 1M | 66K | — |
| llmgateway | $0.03 | $0.13 | $0.006 | $0.0375 | 1M | 1M | — |
| llmgateway-providers | $0.03 | $0.13 | $0.006 | $0.0375 | 1M | 66K | — |
| nano-gpt | $0.03 | $0.13 | $0.006 | $0.038 | 992K | 66K | — |
| nano-gpt · thinking | $0.03 | $0.13 | $0.006 | $0.038 | 984K | 66K | — |
| openrouter | $0.03 | $0.13 | $0.006 | $0.038 | 1M | 66K | — |
| orcarouter | $0.03 | $0.13 | $0.006 | $0.038 | 1M | 66K | — |
| vercel | $0.03 | $0.13 | $0.006 | $0.038 | 991K | 64K | — |
| crossmodel | $0.04 | $0.13 | $0.01 | $0.04 | 1M | 66K | — |
| hyper | $0.20 | $0.80 | $0.04 | — | 1M | 64K | — |
Price history
append-only observations · a listing writes a row only when its price moves
| Date | Host | Field | $/M |
|---|---|---|---|
| 2026-09-02 | openrouter | Cache read | $0.006 |
| 2026-09-06 | openrouter | Cache read | $0.006 |
| 2026-09-02 | openrouter | Cache write | $0.038 |
| 2026-09-06 | openrouter | Cache write | $0.038 |
| 2026-09-02 | openrouter | Input | $0.03 |
| 2026-09-06 | openrouter | Input | $0.03 |
| 2026-09-02 | openrouter | Output | $0.13 |
| 2026-09-06 | openrouter | Output | $0.13 |
More from Alibaba
most-hosted first