Data as of October 19, 2026 · How the score is built
Released Apr 21, 2026262K context
Qwen3.6-27B
Decision readingQwen3.6-27B scores 48.6 out of 100 and ranks #108 of 226. This profile shows 40 source-displayable benchmark rows; its strongest eligible category is Mathematics at #8. Published weights can be self-hosted, but infrastructure cost varies and is not a comparable API token rate.
Released Apr 21, 2026 — see all recent releases
Decision snapshot
Each value carries a field reference instead of floating alone. Markers compare this model with the current ranked and priced catalog; they are not absolute quality thresholds.
Capability
48.6/100
field median 50#108 of 226 ranked models
Public
#108of 226
Verified —
Price
Self-hosted; infrastructure cost varies
input median $0.75No comparable first-party hosted token rate
Speed
56tok/s
field median 94 tok/sFirst token 105.68 s
Context
262Ktokens
field median 256,000Reported for this model; direct source link not stored
Strongest published evidence
Mathematics ranks #8. Particularly strong for mathematical reasoning, scientific computing, and quantitative analysis.
Validate before choosing
40 published rows leave some tracked benchmark slots empty. No comparable first-party API token rate is published.
Source-linked · 40 displayable benchmark rows
Category score record
Scores and ranks appear only where published evidence can be displayed. The table keeps the score, weight, cohort, and evidence state together.
| Category | Score | Rank | Percentile | Weight | Benchmarks | Evidence |
|---|---|---|---|---|---|---|
| AgenticRank #97 of 130Percentile 26thWeight 22%6 benchmarksVerified | 29.1 | 6 benchmarks | Verified | |||
| CodingRank #80 of 151Percentile 47thWeight 20%7 benchmarksVerified | 34.8 | 7 benchmarks | Verified | |||
| ReasoningWeight 17%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| MultimodalRank #39 of 55Percentile 30thWeight 12%15 benchmarksVerified | 67.3 | 15 benchmarks | Verified | |||
| KnowledgeRank #92 of 186Percentile 51stWeight 12%6 benchmarksVerified | 46.8 | 6 benchmarks | Verified | |||
| MultilingualWeight 7%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| Inst. FollowingWeight 5%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| MathRank #8 of 12Percentile 36thWeight 5%6 benchmarksVerified | 65.9 | 6 benchmarks | Verified |
40 of 686 tracked benchmark slots have displayable evidence · bars run 0–100
Coverage detailsHow much of this is verified
Coverage is split by category so a strong number never hides a thin evidence base. Verified means the row is tied to a published source; provisional rows remain visible but separate.
- Agentic6/6 verified
- Coding7/7 verified
- ReasoningNot measured
- Multimodal15/15 verified
- Knowledge6/6 verified
- MultilingualNot measured
- Inst. FollowingNot measured
- Math6/6 verified
Capability shape
Each axis shows percentile within that category’s eligible cohort. The comparison outline is the median of the six nearest public-score peers; a collapsed vertex means the category is not rank-eligible.
Qwen3.6-27B category percentile values
- Agentic26th percentile
- Coding47th percentile
- ReasoningNot eligible
- Multimodal30th percentile
- Knowledge51st percentile
- MultilingualNot eligible
- Instruction followingNot eligible
- Math36th percentile
The dashed outline is median of 6 nearest peers.
Eligible category ranks
- Agentic#97/130
- Coding#80/151
- ReasoningNot ranked
- Multimodal#39/55
- Knowledge#92/186
- MultilingualNot ranked
- Inst. FollowingNot ranked
- Math#8/12
Benchmark ledger
Coding opens by default. The marker compares each value with the best source-verified result in the catalog; provisional leaders do not set the reference. Expand the remaining categories for every published row.
Coding7 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| SWE-bench Pro | Score53.5% | Versus best verified row Best verified: Claude Opus 5.5 · 89.9% | Gap36.4 behind | Weight26% ref. weight | Provider exact |
| SWE Multilingual | Score71.3% | Versus best verified row Best verified: Claude Opus 5.5 · 93.9% | Gap22.6 behind | Weight5% ref. weight | Provider exact |
| Terminal-Bench 2.0 | Score59.3% | Versus best verified row Best verified: GPT-5.5 · 82.0% | Gap22.7 behind | WeightScored in Agentic | Provider exact |
| SWE-bench VerifiedSoftware Engineering Benchmark Verified | Score77.2% | Versus best verified row Best verified: Claude Opus 5 · 96% | Gap18.8 behind | WeightDisplay only | Provider exact |
| LiveCodeBench v6 | Score83.9% | Versus best verified row Best verified: Sakana Fugu-Ultra · 93.2% | Gap9.3 behind | WeightDisplay only | Provider exact |
| NL2Repo | Score36.2% | Versus best verified row Best verified: DeepSeek V4.1 Flash 1017 · 68.2% | Gap32 behind | WeightDisplay only | Provider exact |
| SWE-bench (Vals)SWE-bench, Vals AI run | Score70.0% | Versus best verified row Best verified: Claude Opus 5 · 97.0% | Gap27 behind | WeightDisplay only | Verified |
Agentic6 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| Terminal-Bench 2.0 | Score59.3% | Versus best verified row Best verified: GPT-5.5 · 82% | Gap22.7 behind | Weight7% ref. weight | Provider exact |
| Claw-Eval | Score72.4% | Versus best verified row Best verified: Ornith-1.5-397B · 81.4% | Gap9 behind | WeightDisplay only | Provider exact |
| QwenClawBench | Score53.4% | Versus best verified row Best verified: Qwen3.7 Max · 64.3% | Gap10.9 behind | WeightDisplay only | Provider exact |
| QwenWebBench | Score1487 | Versus best verified row Best verified: Qwen3.7 Max · 1568 | Gap81 behind | WeightDisplay only | Provider exact |
| AndroidWorld | Score70.3% | Versus best verified row Best verified: Qwen3.8-Omni-Flash · 87.1% | Gap16.8 behind | WeightDisplay only | Provider exact |
| Gert LabsGert Labs Composite Game Benchmark | Score54.84% | Versus best verified row Best verified: Claude Opus 4.8 · 72.97% | Gap18.1 behind | WeightDisplay only | Benchmark exact |
Multimodal15 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| MMMU-ProMassive Multi-discipline Multimodal Understanding Pro | Score75.8% | Versus best verified row Best verified: Gemini 3.5 Flash · 83.6% | Gap7.8 behind | WeightWeighted 40% | Provider exact |
| CharXivCharXiv Reasoning | Score78.4% | Versus best verified row Best verified: Qwen3.8 Max · 93.5% | Gap15.1 behind | WeightWeighted 20% | Provider exact |
| MMMUMassive Multi-discipline Multimodal Understanding | Score82.9% | Versus best verified row Best verified: Qwen3.5 397B · 85.0% | Gap2.1 behind | WeightDisplay only | Provider exact |
| RealWorldQA | Score84.1% | Versus best verified row Best verified: Qwen3.8-Flash-Next · 88.5% | Gap4.4 behind | WeightDisplay only | Provider exact |
| DynaMath | Score85.6% | Versus best verified row Best verified: Qwen3.5 397B · 86.3% | Gap0.7 behind | WeightDisplay only | Provider exact |
| MStar | Score81.4% | Versus best verified row Best verified: Qwen3.6-27B · 81.4% | GapBest verified | WeightDisplay only | Provider exact |
| SimpleVQA | Score56.1% | Versus best verified row Best verified: Qwen3.7 Plus · 81.7% | Gap25.6 behind | WeightDisplay only | Provider exact |
| CC-OCR | Score81.2% | Versus best verified row Best verified: Qwen3.5 397B · 82.0% | Gap0.8 behind | WeightDisplay only | Provider exact |
| CountBench | Score97.8% | Versus best verified row Best verified: Qwen3.6-27B · 97.8% | GapBest verified | WeightDisplay only | Provider exact |
| RefCOCO (avg)RefCOCO average | Score92.5% | Versus best verified row Best verified: Qwen3.6-27B · 92.5% | GapBest verified | WeightDisplay only | Provider exact |
| ERQA | Score62.5% | Versus best verified row Best verified: Qwen3.8 Max · 77.8% | Gap15.3 behind | WeightDisplay only | Provider exact |
| Video-MME (with subtitle)Video-MME with subtitle | Score87.7% | Versus best verified row Best verified: Qwen3.8 Max · 90.4% | Gap2.7 behind | WeightDisplay only | Provider exact |
| VideoMMMU | Score84.4% | Versus best verified row Best verified: Qwen3.8 Max · 88.7% | Gap4.3 behind | WeightDisplay only | Provider exact |
| MLVU (M-Avg)MLVU mean average | Score86.6% | Versus best verified row Best verified: Qwen3.8 Max · 90.8% | Gap4.2 behind | WeightDisplay only | Provider exact |
| V* | Score94.7% | Versus best verified row Best verified: Qwen3.6 Plus · 96.9% | Gap2.2 behind | WeightDisplay only | Provider exact |
Knowledge6 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| HLEHumanity's Last Exam | Score24% | Versus best verified row Best verified: Claude Fable 5.1 · 65% | Gap41 behind | Weight44% ref. weight | Provider exact |
| MMLU-ProMassive Multitask Language Understanding Professional | Score86.2% | Versus best verified row Best verified: Qwen3.7 Max · 89.6% | Gap3.4 behind | Weight6% ref. weight | Provider exact |
| GPQAGraduate-Level Google-Proof Q&A | Score87.8% | Versus best verified row Best verified: GPT-6 Astra · 96% | Gap8.2 behind | Weight3% ref. weight | Provider exact |
| SuperGPQASuperGPQA: Scaling LLM Evaluation Across 285 Graduate Disciplines | Score66% | Versus best verified row Best verified: Qwen 3.6 Max (preview) · 73.9% | Gap7.9 behind | Weight2% ref. weight | Provider exact |
| MMLU-Redux | Score93.5% | Versus best verified row Best verified: Qwen3.7 Max · 95% | Gap1.5 behind | WeightDisplay only | Provider exact |
| C-Eval | Score91.4% | Versus best verified row Best verified: Qwen3.6 Plus · 93.3% | Gap1.9 behind | WeightDisplay only | Provider exact |
Math6 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| FrontierMath v2 (Tiers 1-3)FrontierMath v2 Tiers 1-3 | Score35.088% | Versus best verified row Best verified: GPT-6 Astra · 93.684% | Gap58.6 behind | WeightWeighted 30% | Benchmark exact |
| HMMT Feb 2026Harvard-MIT Mathematics Tournament February 2026 | Score84.3% | Versus best verified row Best verified: Qwen3.7 Max · 97.1% | Gap12.8 behind | WeightWeighted 25% | Provider exact |
| AIME26AIME 2026 | Score94.1% | Versus best verified row Best verified: GLM-5.2 · 99.2% | Gap5.1 behind | WeightWeighted 25% | Provider exact |
| HMMT Feb 2025Harvard-MIT Mathematics Tournament February 2025 | Score93.8% | Versus best verified row Best verified: GLM-4.7 · 97.1% | Gap3.3 behind | WeightDisplay only | Provider exact |
| HMMT Nov 2025Harvard-MIT Mathematics Tournament November 2025 | Score90.7% | Versus best verified row Best verified: Qwen3.6 Plus · 94.6% | Gap3.9 behind | WeightDisplay only | Provider exact |
| MMAnswerBench | Score80.8% | Versus best verified row Best verified: GLM-5.2 · 91.0% | Gap10.2 behind | WeightDisplay only | Provider exact |
Lineage
The sequence follows explicit supersedes links. Each score is estimated for that model; a relative can inform a sparse estimate but never sets a floor, so a newer release can score below an earlier one. Scores and prices remain blank when the corresponding public row or first-party rate is unavailable.
Base entry
Spec sheet
Each documented value carries its source. Missing fields stay visible as not sourced or not published, rather than disappearing from the page.
- API model ID
- Not published
- Context window
- 262K
- Maximum output
- Not sourced yet
- Knowledge cutoff
- Not sourced yet
- Input modalities
- Not sourced yet
- Output modalities
- Not sourced yet
- Parameters
- Not sourced yet
- Availability
- Not sourced yet
- Cloud regions
- Not tracked yet
- Lifecycle
- Superseded
- API capabilities
- Tool calling, structured outputs, and batch support are not tracked yet
- Prompt caching
- Not documented in the pricing record
- Self-host
- Qwen3.6-27B needs ~24GB VRAM (1× NVIDIA RTX 4090 (24GB)).
- Rate limits
- Not tracked yet
How to read this profile
The visual layer above carries the decisions. These notes preserve the model, ranking, coverage, and family context behind the numbers.
Qwen3.6-27B ranks #108 of 226 on the public leaderboard with a score of 48.57/100. It does not yet have enough sourced coverage for a verified position.
Qwen3.6-27B is a open weight model with a 262K context window. It uses an explicit reasoning mode, which can improve complex problem solving while adding latency and token use.
Official exact-value snapshot from the Qwen/Qwen3.6-27B Hugging Face model card. The public repository and FP8 checkpoint were created on April 21, 2026, and the model card states a native 262,144-token context window extensible to 1,010,000 tokens.
40 of 686 tracked benchmark slots currently have displayable evidence. Missing categories stay blank.
Its strongest eligible category is Mathematics at #8, while its lowest eligible position is Agentic at #97. particularly strong for mathematical reasoning, scientific computing, and quantitative analysis.
Last updated October 19, 2026. Runtime fields remain blank until a sourced snapshot exists.
Deployment options
Self-host and provider-specific paths stay separate from benchmark evidence so operating constraints are visible before a score becomes the whole decision.
Self-host vs API cost
Estimates at 50,000 req/day · 1000 tokens/req average.
Questions
How does Qwen3.6-27B perform overall in AI benchmarks?
Qwen3.6-27B ranks #108 out of 226 models on the public BenchAlign leaderboard, with a score of 48.57/100. Its evidence status is Estimated, and this profile shows 40 source-displayable benchmark rows. The label describes evidence depth, not a provider quality claim; inspect category rows before choosing a workload.
Is Qwen3.6-27B good for knowledge and understanding?
Qwen3.6-27B ranks #92 out of 186 eligible models for knowledge and understanding, with a public category score of 46.8/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Qwen3.6-27B good for coding and programming?
Qwen3.6-27B ranks #80 out of 151 eligible models for coding and programming, with a public category score of 34.8/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Qwen3.6-27B good for mathematics?
Qwen3.6-27B ranks #8 out of 12 eligible models for mathematics, with a public category score of 65.9/100. That places it in the current top ten for this category. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Qwen3.6-27B good for agentic tool use and computer tasks?
Qwen3.6-27B ranks #97 out of 130 eligible models for agentic tool use and computer tasks, with a public category score of 29.1/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Qwen3.6-27B good for multimodal and grounded tasks?
Qwen3.6-27B ranks #39 out of 55 eligible models for multimodal and grounded tasks, with a public category score of 67.3/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Qwen3.6-27B open source?
Qwen3.6-27B is an open-weight model from Alibaba. Its weights can be downloaded for local or hosted deployment, subject to the published license. Open weight does not automatically mean open source: training data and training code may remain private, and commercial restrictions can still apply.
Does Qwen3.6-27B have full benchmark coverage on BenchLM?
No. Qwen3.6-27B currently has 56 source-displayable rows across 686 tracked benchmark slots. The profile exposes published, non-generated evidence and leaves missing categories blank until an exact evaluation is available. Coverage describes how much was measured; it is not a penalty added to an individual benchmark result.
What is the context window size of Qwen3.6-27B?
Qwen3.6-27B has a reported context window of 262K in the exact-model catalog record. The value stays visible, but the profile marks its source link as unavailable instead of presenting it as directly documented. Maximum output length remains separate because providers often publish a different limit.