Alibaba Cloud / Qwenopen-weight
Released: 2025-03-05QwQ 32B
Alibaba 32B parameter open-weight reasoning model matching frontier model math and coding capabilities.
Intelligence Index
86.1/ 100
Calibrated multi-domain compositeThroughput Speed
45tok/sec
Output streaming throughputTime To First Token
600ms
Initial chunk server latencyLiveBench Contam-Free
73.8%
September 2026 suite scoreLiveBench Multi-Domain Evaluation Breakdown
Monthly refreshed contamination-free test setsLogical Reasoning86.2%
Coding & Repo Repair74.8%
Mathematics (AIME)89.1%
Data Analysis & Tables75.2%
Language & Comprehension79.5%
Instruction Following85.0%
Sourced Benchmark Evaluations (61)
Verified performance across authoritative benchmarks with provenance tracking.| Benchmark | Status | Score | Trust | Date | Source Type | Provenance |
|---|---|---|---|---|---|---|
| ARC-AGI-3 | Active | 56% | 93 | 2025-04-09 | independent | Source ↗ |
| ARC-AGI-2 | Nearing Saturation | 67.2% | 75 | 2025-04-02 | independent | Source ↗ |
| FrontierMath | Active | 24.9% | 98 | 2025-03-30 | independent | Source ↗ |
| PutnamBench | Active | 46.3% | 90 | 2025-03-29 | independent | Source ↗ |
| SWE-Lancer | Deprecated | 50.9% | 20 | 2025-03-27 | independent | Source ↗ |
| Omni-MATH | Deprecated | 65.9% | 10 | 2025-03-27 | independent | Source ↗ |
| Soohak | Active | 67.7% | 98 | 2025-03-25 | independent | Source ↗ |
| FrontierCode | Active | 43.4% | 98 | 2025-03-25 | independent | Source ↗ |
| Humanity's Last Exam | Active | 37.9% | 93 | 2025-03-23 | independent | Source ↗ |
| StructFlowBench | Nearing Saturation | 72.3% | 75 | 2025-03-23 | independent | Source ↗ |
| AIME | Nearing Saturation | 84.6% | 75 | 2025-03-23 | independent | Source ↗ |
| OlympiadBench | Active | 73.1% | 80 | 2025-03-22 | independent | Source ↗ |
| MathArena | Active | 71.3% | 99 | 2025-03-21 | independent | Source ↗ |
| FireBench | Active | 71.4% | 95 | 2025-03-21 | independent | Source ↗ |
| Terminal-Bench | Active | 52.4% | 80 | 2025-03-21 | independent | Source ↗ |
| OpenAI-MRCR | Active | 72.3% | 80 | 2025-03-20 | independent | Source ↗ |
| SimpleQA | Active | 44.8% | 90 | 2025-03-20 | independent | Source ↗ |
| CodeContests | Deprecated | 48.6% | 10 | 2025-03-20 | independent | Source ↗ |
| ComplexBench | Nearing Saturation | 69.7% | 80 | 2025-03-19 | independent | Source ↗ |
| SWE-bench Pro | Active | 46.4% | 90 | 2025-03-19 | independent | Source ↗ |
| LongBench-Pro | Active | 70.6% | 80 | 2025-03-19 | independent | Source ↗ |
| ZeroSCROLLS | Nearing Saturation | 68.9% | 78 | 2025-03-18 | independent | Source ↗ |
| MultiPL-E | Deprecated | 68.8% | 10 | 2025-03-18 | independent | Source ↗ |
| LiveCodeBench | Saturated | 61.3% | 39 | 2025-03-17 | independent | Source ↗ |
| AgentIF | Nearing Saturation | 57.8% | 80 | 2025-03-17 | independent | Source ↗ |
| MATH | Saturated | 87.3% | 25 | 2025-03-17 | independent | Source ↗ |
| InfiniteBench | Nearing Saturation | 73.2% | 65 | 2025-03-17 | independent | Source ↗ |
| Mercury | Deprecated | 65.8% | 10 | 2025-03-16 | independent | Source ↗ |
| SCROLLS | Nearing Saturation | 68.9% | 65 | 2025-03-16 | independent | Source ↗ |
| MMLU-Pro | Nearing Saturation | 79.2% | 75 | 2025-03-16 | independent | Source ↗ |
| RULER | Nearing Saturation | 79.2% | 84 | 2025-03-15 | independent | Source ↗ |
| TabMWP | Saturated | 82% | 25 | 2025-03-15 | independent | Source ↗ |
| CyberSecEval | Deprecated | 67.3% | 10 | 2025-03-15 | independent | Source ↗ |
| SWE-bench Verified | Saturated | 58.3% | 25 | 2025-03-14 | independent | Source ↗ |
| MMLU-Redux | Saturated | 84.4% | 25 | 2025-03-14 | independent | Source ↗ |
| QuALITY | Nearing Saturation | 75.8% | 78 | 2025-03-14 | independent | Source ↗ |
| LongBench | Nearing Saturation | 74% | 65 | 2025-03-13 | independent | Source ↗ |
| SWE-bench | Deprecated | 50.9% | 10 | 2025-03-13 | independent | Source ↗ |
| MGSM | Nearing Saturation | 84.6% | 65 | 2025-03-13 | independent | Source ↗ |
| AGIEval | Saturated | 74.1% | 25 | 2025-03-12 | independent | Source ↗ |
| QASPER | Nearing Saturation | 71.5% | 65 | 2025-03-12 | independent | Source ↗ |
| TruthfulQA | Saturated | 75.8% | 25 | 2025-03-12 | independent | Source ↗ |
| GSM-Hard | Deprecated | 85.5% | 10 | 2025-03-11 | independent | Source ↗ |
| SuperGLUE | Saturated | 87.9% | 25 | 2025-03-11 | independent | Source ↗ |
| CodeXGLUE | Deprecated | 70.3% | 23 | 2025-03-10 | independent | Source ↗ |
| MT-Bench | Nearing Saturation | 8.8% | 75 | 2025-03-10 | independent | Source ↗ |
| BIG-Bench Hard | Saturated | 84.5% | 25 | 2025-03-10 | independent | Source ↗ |
| WinoGrande | Saturated | 89.6% | 28 | 2025-03-09 | independent | Source ↗ |
| MBPP+ | Nearing Saturation | 76.3% | 65 | 2025-03-09 | independent | Source ↗ |
| IFEval | Deprecated | 83.3% | 25 | 2025-03-09 | independent | Source ↗ |
| GPQA | Nearing Saturation | 75.8% | 78 | 2025-03-09 | independent | Source ↗ |
| Needle-in-a-Haystack | Saturated | 97.2% | 44 | 2025-03-08 | independent | Source ↗ |
| HumanEval+ | Saturated | 80.8% | 25 | 2025-03-08 | independent | Source ↗ |
| CommonsenseQA | Saturated | 90.5% | 28 | 2025-03-08 | independent | Source ↗ |
| MBPP | Saturated | 79.3% | 25 | 2025-03-07 | independent | Source ↗ |
| GSM8K | Saturated | 96.2% | 25 | 2025-03-07 | independent | Source ↗ |
| GLUE | Saturated | 93.1% | 28 | 2025-03-07 | independent | Source ↗ |
| MMLU | Saturated | 87.8% | 25 | 2025-03-06 | independent | Source ↗ |
| HumanEval | Saturated | 85.3% | 25 | 2025-03-06 | independent | Source ↗ |
| ARC-AGI | Saturated | 81% | 25 | 2025-03-06 | independent | Source ↗ |
| LiveBench (Continuous) | Active | 73.8% | 94 | 2025-02-01 | independent | Source ↗ |
Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run qwq-32b
# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "qwq-32b",
"messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
}'Evaluation & Sourcing Notes
Scores listed for QwQ 32B represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.
All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.