AlibabaAuto-detectedQwen3.5-4B
Best configuration ranks #356 of 610 on the BenchLeader Index at 47.1 ±6.6. Released 24 Feb 2026.
- Blended price
- $0.060/M
- $0.030 in · $0.150 out
- Output speed
- 27 tok/s
- First answer
- 74 s
- Context
- 262k
- Overall index47
- Agents & tools49
- Maths31
- Knowledge36
- Instruction following53
- Multimodal50
- Long context58
- Composite46
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 | 43.9 | #429 | 22 tok/s | 0.80 s | $0.0001 | Agents & tools 43 · Composite 43 · Instruction following 37 · Knowledge 29 · Long context 43 · Maths 49 · Multimodal 46 |
| defaultbest | 47.1 | #356 | 27 tok/s | 74 s | $0.0001 | Agents & tools 49 · Composite 46 · Instruction following 53 · Knowledge 36 · Long context 58 · Maths 31 · Multimodal 50 |
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 (AA)not in index | 71.2%#278 | 77.1%#217 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 8.0%#284 | 9.9%#259 | Artificial Analysis |
Agents & tools
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| Terminal-Bench Hard | 11.4%#207 | 18.2%#166 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 87.7%#76 | 92.1%#53 | Artificial Analysis |
Maths
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| OTIS Mock AIME | 55.8%#156 | – | Epoch AI Benchmarking Hub |
| AIME 2026 | – | 89.7%#28 | MathArena |
| HMMT February 2026 | – | 72.9%#28 | MathArena |
Knowledge
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| AA-Omniscience | -74.7#448 | -58.4#377 | Artificial Analysis |
Instruction following
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| IFBench | 33.3%#315 | 52.0%#161 | Artificial Analysis |
Multimodal
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| MMMU-Pro | 62.1%#166 | 65.4%#142 | Artificial Analysis |
Long context
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| AA-LCR | 34.7%#305 | 63.0%#206 | Artificial Analysis |
Composite
| Benchmark | no reasoning | default | Source |
|---|---|---|---|
| AA Intelligence Index | 10.8#302 | 13.1#251 | Artificial Analysis |
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 | $0.0001 | – | 1.4 min |
| Summarise a 30-page report | 12,000 / 600 | $0.0005 | – | 1.6 min |
| Code edit | 6,000 / 1,500 | $0.0004 | – | 2.1 min |
| Agentic coding session | 60,000 / 4,000 | $0.0024 | – | 3.7 min |
| Structured extraction | 2,000 / 200 | $0.0001 | – | 1.4 min |
See also
Data as of 9 Sept 2026. Compare these configurations.