Qwen3.5 4B is a compact open-weight multimodal model from Alibaba for reasoning, coding, visual understanding, tool use, and structured output.
Specification
most-agreed values
Context
262K
Max output
33K
Released
2026-08-16
Knowledge cutoff
2025-04
Retires
—
Open weights
yes
Input
text, image
Output
text
Price
US dollars per million tokens · most-agreed
Input
$0.10
Output
$0.20
Cache read
$0.05
Quality
1 benchmark · 1 source
Composite
60th
percentile of 334 scored models
Rank
134 / 334
±0.14 sd
Evidence
1 × 1
too little evidence to place with confidence
Effort range
—
one configuration only
math60th1/7 bench
Every score
one row per source and configuration — nothing averaged away
| Benchmark | Score | Configuration | Source | Run |
|---|---|---|---|---|
| OTIS Mock AIME 2024-2025 | 55.8 ±6.2 | default | Epoch AI | 2026-08-28 |
The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.
Available from 4 hosts
| Host | In $/M | Out $/M | Cache rd | Cache wr | Context | Output | Retires |
|---|---|---|---|---|---|---|---|
| qvac | free | free | — | — | 33K | 8K | — |
| siliconflow-cn | free | free | — | — | 262K | 66K | — |
| empiriolabs | $0.04 | $0.07 | $0.02 | — | 262K | 33K | — |
| nano-gpt | $0.10 | $0.20 | $0.05 | — | 262K | 33K | — |
More from Alibaba
most-hosted first