AlibabaQwen3.5 27B
Best configuration ranks #177 of 610 on the BenchLeader Index at 55.4 ±5.3. Last measured 8 Sept 2026. Released 24 Feb 2026.
- Blended price
- $0.825/M
- $0.300 in · $2.40 out
- Output speed
- 76 tok/s
- First answer
- 32 s
- first token 0.90 s
- Context
- 262k
- Overall index55
- Reasoning59
- Coding46
- Agents & tools62
- Maths41
- Knowledge43
- Instruction following73
- Human preference59
- Multimodal59
- Long context65
- Composite58
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 | 51.9 | #262 | 81 tok/s | 5.65 s | $0.0008 | Agents & tools 61 · Composite 53 · Instruction following 48 · Knowledge 42 · Long context 58 · Multimodal 54 |
| defaultbest | 55.4 | #177 | 76 tok/s | 32 s | $0.0008 | Agents & tools 62 · Coding 46 · Composite 58 · Human preference 59 · Instruction following 73 · Knowledge 43 · Long context 65 · Maths 41 · Multimodal 59 · Reasoning 59 |
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 | Trend |
|---|---|---|---|---|
| LMArena Hard Prompts | – | 1427#122 | LMArena | |
| GPQA Diamond (AA)not in index | 84.2%#144 | 85.8%#118 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 13.9%#202 | 23.9%#139 | Artificial Analysis |
Coding
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| WeirdML | – | 39.5%#103 | WeirdML | |
| LMArena Coding | – | 1450#128 | LMArena | |
| LMArena WebDev | – | 1358#88 | LMArena |
Agents & tools
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| Terminal-Bench Hard | 31.8%#99 | 32.6%#95 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 87.1%#79 | 93.9%#39 | Artificial Analysis |
Maths
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| AIME 2026 | – | 91.7%#24 | MathArena | |
| HMMT February 2026 | – | 81.1%#26 | MathArena | |
| MathArena Apex | – | 2.1%#29 | MathArena |
Knowledge
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| AA-Omniscience | -47.7#297 | -44.0#279 | Artificial Analysis |
Instruction following
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| IFBench | 46.9%#189 | 75.6%#29 | Artificial Analysis |
Human preference
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| LMArena Text | – | 1408#124 | LMArena |
Multimodal
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| LMArena Vision | – | 1242#56 | LMArena | |
| MMMU-Pro | 70.0%#118 | 75.0%#76 | Artificial Analysis |
Long context
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| AA-LCR | 64.0%#199 | 77.7%#87 | Artificial Analysis |
Composite
| Benchmark | no reasoning | default | Source | Trend |
|---|---|---|---|---|
| AA Intelligence Index | 19.4#181 | 22.9#137 | 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. | 53 tok/s | 0.59 s | $0.195 | $1.56 | 262k | – |
| DeepInfra | 46 tok/s | 0.32 s | $0.260 | $2.60 | 262k | fp8 |
| AtlasCloud | 17 tok/s | 0.93 s | $0.270 | $2.16 | 262k | fp8 |
| Phala | 15 tok/s | 2.97 s | $0.300 | $2.40 | 262k | – |
| NovitaAI | 15 tok/s | 0.87 s | $0.300 | $2.40 | 262k | bf16 |
| SiliconFlow | 11 tok/s | 1.26 s | $0.250 | $2.00 | 262k | fp8 |
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. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0008 | – | 36.0 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0050 | – | 40.0 s |
| Code edit | 6,000 / 1,500 | $0.0054 | – | 51.9 s |
| Agentic coding session | 60,000 / 4,000 | $0.028 | – | 1.4 min |
| Structured extraction | 2,000 / 200 | $0.0011 | – | 34.7 s |
See also
Data as of 9 Sept 2026. Compare these configurations.