Qwen3.5-27B
Qwen3.5-27BAlibabaOpen weightsReasoningProvisionalReleased Mar 4, 2026Input: text, image, video
Overall score
48.6
Rank 107 of 222 · 38.9–58.4
Context window
262K
262,000 tokens
Price per 1M tokens
$0.27 in · $2.16 out
Source: OpenRouter
Blended price
$0.74
3:1 input:output tokens
Output speed
—
Not measured yet
Category scores
Scores per capability area with the model's rank among ranked models in that category.
- 1Instruction following#1492.9
- 2Multilingual#1087.1
- 3Knowledge#9770.4
- 4Reasoning70.2
- 5Multimodal#3867.5
- 6Coding61.6
- 7Agentic & tool use#7619.7
Benchmark results (19)
As reported in system cards, papers and public evaluations. Position compares this model with every other model reporting the same benchmark.
Multimodal
| MMMU-Pro | 75.0 | #34 of 48 |
| CharXiv | 79.5 | #29 of 42 |
| MMMU | 82.3 | #8 of 21 |
| MathVision | 86.0 | #14 of 20 |
| V* | 93.7 | #6 of 12 |
| MMVU | 73.3 | #6 of 7 |
Agentic & tool use
| Gert Labs | 39.41 | #39 of 52 |
| BrowseComp | 61.0 | #39 of 50 |
| Terminal-Bench 2.0 | 41.6 | #36 of 39 |
| OSWorld-Verified | 56.2 | #32 of 35 |
Coding
| SWE-bench Verified | 72.4 | #54 of 98 |
| LiveCodeBench v6 | 80.7 | #23 of 46 |
| SWE-Rebench | 58.9 | #6 of 13 |
Knowledge
| GPQA | 85.5 | #48 of 117 |
| MMLU-Pro | 86.1 | #15 of 67 |
| SuperGPQA | 65.6 | #14 of 20 |
Reasoning
| LongBench v2 | 60.6 | #11 of 17 |
Multilingual
| MMLU-ProX | 82.2 | #9 of 16 |
Instruction following
| IFEval | 95.0 | #2 of 53 |
Pricing by source
USD per 1M tokens.
| Source | Input | Output | Cached in |
|---|---|---|---|
| List price (OpenRouter) | $0.27 | $2.16 | — |
| OpenRouter qwen/qwen3.5-27b | $0.27 | $2.16 | $0.27 |
Open-weight model. Self-hosted or third-party hosted costs vary by provider and infrastructure.
Price history
- Oct 17, 2026: $0.26 / $2.60
- Oct 19, 2026: $0.27 / $2.16
Arena ratings
Human-preference ratings from LMArena. Best-rated variant shown.
| Text | 1409 | #142 · 28.7K votes |
| WebDev (coding) | 1358 | #114 · 7.8K votes |
| Vision | 1241 | #72 · 30.3K votes |
Open weights
From the Hugging Face repository.
- Parameters
- 28B
- Licence
- apache-2.0
- Downloads (30 days)
- 1.8M
- Repository
- Qwen/Qwen3.5-27B ↗
Speed
Measured by us through OpenRouter (median of trials).
Not measured yet.
Related models
Similar score
- Qwen3.6-27B48.6 · $0.69
- Quasar 438B48.6 · $0.90
- GLM-4.748.8 · $0.90
- Qwen3.5 Plus48.9 · $0.90
- GPT-5 mini48.4 · $0.69