AlibabaQwen3.5-4B
Qwen3.5-4B is an Alibaba open-weights reasoning model, released 24 Feb 2026. Its best configuration (thinking reasoning effort) ranks #456 of 751 on the BenchLeader Index at 46.7 ±5.1, in the lower half. The ± is the point: 312 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in long context (57) and lowest in knowledge (36). At $0.060 per million tokens blended it is among the cheapest fifth of ranked models. Output speed of 22 tokens per second puts it in the slowest quarter, with a first answer in 92.3 s. It has been measured at 3 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them.
- Blended price
- $0.060/M
- $0.030 in · $0.150 out
- Output speed
- 22 tok/s
- measured by Artificial Analysis
- First answer
- 92 s
- Context
- 262k
- Full answer
- 115 s
- median, reasoning included
- Cost per run
- –
- Released
- 24 Feb 2026
- Overall index47
- Reasoning41
- Agents & tools40
- Knowledge36
- Instruction following53
- Multimodal48
- Long context57
- Composite44
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. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoning | 42.4 | #571 | 12 tok/s | 0.94 s | $0.0001 | Agents & tools 36 · Composite 42 · Instruction following 37 · Knowledge 28 · Long context 42 · Maths 48 · Multimodal 45 · Reasoning 41 |
| thinkingbest | 46.7 | #456 | 22 tok/s | 92 s | $0.0001 | Agents & tools 40 · Composite 44 · Instruction following 53 · Knowledge 36 · Long context 57 · Multimodal 48 · Reasoning 41 |
| not stated | – | – | – | – | – | Maths 31 |
How its index has moved
9 Sept 2026 to 3 Oct 2026- thinking
- no reasoning
The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
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 | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| GPQA Diamond (AA)not in index | 71.2%#282 | 77.1%#219 | – | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 8.0%#332 | 9.9%#304 | – | Artificial Analysis | |
| CritPt | 0.0%#293 | 0.0%#293 | – | Artificial Analysis |
Coding
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| Terminal-Bench 2.1 (AA)not in index | 21.4%#166 | 25.8%#163 | – | Artificial Analysis |
Agents & tools
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| Terminal-Bench Hard | 11.4%#212 | 18.2%#170 | – | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 87.7%#77 | 92.1%#53 | – | Artificial Analysis | |
| τ²-Bench Banking (AA) | 4.3%#190 | 6.8%#162 | – | Artificial Analysis |
Maths
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| OTIS Mock AIME | 55.8%#165 | – | – | Epoch AI Benchmarking Hub | |
| AIME 2026 | – | – | 89.7%#28 | MathArena | |
| HMMT February 2026 | – | – | 72.9%#28 | MathArena |
Knowledge
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| AA-Omniscience | -74.7#501 | -58.4#427 | – | Artificial Analysis | |
| AA-Omniscience: accuracynot in index | 11.2%#457 | 15.1%#395 | – | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 3.2%#518 | 13.4%#343 | – | Artificial Analysis |
Instruction following
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| IFBench | 33.3%#324 | 52.0%#162 | – | Artificial Analysis |
Multimodal
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| MMMU-Pro | 62.1%#194 | 65.4%#170 | – | Artificial Analysis |
Long context
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| AA-LCR | 34.7%#354 | 63.0%#249 | – | Artificial Analysis |
Composite
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| AA Intelligence Index | 10.8#350 | 13.1#296 | – | 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 | – | 1.8 min |
| Summarise a 30-page report | 12,000 / 600 | $0.0005 | – | 2.0 min |
| Code edit | 6,000 / 1,500 | $0.0004 | – | 2.7 min |
| Agentic coding session | 60,000 / 4,000 | $0.0024 | – | 4.6 min |
| Structured extraction | 2,000 / 200 | $0.0001 | – | 1.7 min |
See also
Data as of 4 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “Qwen3.5-4B: benchmarks, pricing, speed and rank”, https://www.benchleader.com/models/qwen3-5-4b, data as of 4 Oct 2026.