AlibabaQwen3.6 27B
Best configuration ranks #231 of 610 on the BenchLeader Index at 53.2 ±8.0. Last measured 1 Sept 2026. Released 22 Apr 2026.
- Blended price
- $1.35/M
- $0.600 in · $3.60 out
- Output speed
- 56 tok/s
- First answer
- 105 s
- first token 2.07 s
- Context
- 262k
- Overall index53
- Reasoning62
- Agents & tools64
- Maths53
- Knowledge55
- Instruction following66
- Human preference24
- Multimodal59
- Long context65
- Composite37
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Default” means the publisher did not say which setting was used.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoning | 50.7 | #280 | 57 tok/s | 3.62 s | $0.0013 | Agents & tools 52 · Coding 40 · Composite 54 · Instruction following 47 · Knowledge 39 · Long context 60 · Maths 49 · Multimodal 56 · Reasoning 62 |
| defaultbest | 53.2 | #231 | 56 tok/s | 105 s | $0.0013 | Agents & tools 64 · Composite 37 · Human preference 24 · Instruction following 66 · Knowledge 55 · Long context 65 · Maths 53 · Multimodal 59 · Reasoning 62 |
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| GPQA Diamond | 84.8%#81 | 85.9%#72 | Epoch AI Benchmarking Hub |
| LiveBench Reasoningnot in index | – | 70.3%#51 | LiveBench |
| GPQA Diamond (AA)not in index | 82.9%#160 | 84.2%#144 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 15.1%#193 | 23.1%#144 | Artificial Analysis |
Coding
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| SciCode | 37.3%#123 | – | SciCode |
| LiveBench Codingnot in index | – | 71.8%#44 | LiveBench |
| SciCode (AA)not in index | – | 42.8%#103 | Artificial Analysis |
| SWE-bench (Vals)not in index | – | 70.0%#58 | Vals AI |
Agents & tools
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| LiveBench Agentic Codingnot in index | – | 39.3%#50 | LiveBench |
| Terminal-Bench Hard | 21.2%#152 | 34.9%#80 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 93.6%#42 | 94.2%#34 | Artificial Analysis |
Maths
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| FrontierMath Tiers 1–3 | 34.0%#55 | 35.1%#52 | Epoch AI Benchmarking Hub |
| OTIS Mock AIME | 66.7%#134 | 91.1%#55 | Epoch AI Benchmarking Hub |
| LiveBench Mathematicsnot in index | – | 79.9%#46 | LiveBench |
Knowledge
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| LiveBench Data Analysisnot in index | – | 70.4%#38 | LiveBench |
| AA-Omniscience | -53.2#343 | -20.0#183 | Artificial Analysis |
| CorpFin | – | 62.3%#48 | Vals AI |
| TaxEval | – | 71.3%#77 | Vals AI |
Instruction following
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| LiveBench Languagenot in index | – | 63.3%#51 | LiveBench |
| IFBench | 45.7%#198 | 67.5%#84 | Artificial Analysis |
Human preference
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| EQ-Bench 4 | – | 1026#26 | EQ-Bench |
Multimodal
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| MMMU-Pro | 71.7%#106 | 74.6%#81 | Artificial Analysis |
Long context
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| AA-LCR | 66.7%#186 | 77.3%#91 | Artificial Analysis |
Composite
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| Epoch Capabilities Indexnot in index | – | 146.5#60 | Epoch AI Benchmarking Hub |
| LiveBench | – | 64.0%#50 | LiveBench |
| AA Intelligence Index | 19.8#175 | 21.9#153 | Artificial Analysis |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| Alibaba Cloud Int. | 56 tok/s | 1.59 s | $0.450 | $2.70 | 262k | – |
| Venice | 47 tok/s | 0.82 s | $0.325 | $3.25 | 256k | fp8 |
| DeepInfra | 41 tok/s | 1.07 s | $0.320 | $3.20 | 262k | fp8 |
| SiliconFlow | 18 tok/s | 2.54 s | $0.300 | $3.20 | 262k | fp8 |
| Chutes | 14 tok/s | 3.46 s | $0.300 | $2.00 | 262k | fp8 |
| Phala | 10 tok/s | 4.47 s | $0.320 | $2.70 | 262k | – |
Price history
Listed price per 1M tokens over time, as recorded by OpenRouter for the provider with the longest history.
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.300 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0013 | $0.0012 | 1.8 min |
| Summarise a 30-page report | 12,000 / 600 | $0.0094 | $0.0067 | 1.9 min |
| Code edit | 6,000 / 1,500 | $0.0090 | $0.0077 | 2.2 min |
| Agentic coding session | 60,000 / 4,000 | $0.050 | $0.037 | 2.9 min |
| Structured extraction | 2,000 / 200 | $0.0019 | $0.0015 | 1.8 min |
See also
Data as of 9 Sept 2026. Compare these configurations.