Alibaba Cloud / Qwenopen-weight
Released: 2026-06-18Qwen3-235B
Alibaba 235B open-weight Mixture-of-Experts foundation model supporting dual thinking/non-thinking inference modes.
Intelligence Index
91.5/ 100
Calibrated multi-domain compositeThroughput Speed
70tok/sec
Output streaming throughputTime To First Token
410ms
Initial chunk server latencyLiveBench Contam-Free
77.4%
September 2026 suite scoreLiveBench Multi-Domain Evaluation Breakdown
Monthly refreshed contamination-free test setsLogical Reasoning87.5%
Coding & Repo Repair80.5%
Mathematics (AIME)91.0%
Data Analysis & Tables74.8%
Language & Comprehension70.5%
Instruction Following64.0%
Sourced Benchmark Evaluations (61)
Verified performance across authoritative benchmarks with provenance tracking.| Benchmark | Status | Score | Trust | Date | Source Type | Provenance |
|---|---|---|---|---|---|---|
| ARC-AGI-3 | Active | 59.5% | 93 | 2026-07-23 | independent | Source ↗ |
| ARC-AGI-2 | Nearing Saturation | 71.4% | 75 | 2026-07-16 | independent | Source ↗ |
| FrontierMath | Active | 25.5% | 98 | 2026-07-13 | independent | Source ↗ |
| PutnamBench | Active | 47.3% | 90 | 2026-07-12 | independent | Source ↗ |
| SWE-Lancer | Deprecated | 54.7% | 20 | 2026-07-10 | independent | Source ↗ |
| Omni-MATH | Deprecated | 67.3% | 10 | 2026-07-10 | independent | Source ↗ |
| Soohak | Active | 69.2% | 98 | 2026-07-08 | independent | Source ↗ |
| FrontierCode | Active | 46.7% | 98 | 2026-07-08 | independent | Source ↗ |
| Humanity's Last Exam | Active | 40.3% | 93 | 2026-07-06 | independent | Source ↗ |
| StructFlowBench | Nearing Saturation | 54.4% | 75 | 2026-07-06 | independent | Source ↗ |
| AIME | Nearing Saturation | 86.5% | 75 | 2026-07-06 | independent | Source ↗ |
| OlympiadBench | Active | 74.6% | 80 | 2026-07-05 | independent | Source ↗ |
| MathArena | Active | 72.8% | 99 | 2026-07-04 | independent | Source ↗ |
| FireBench | Active | 53.8% | 95 | 2026-07-04 | independent | Source ↗ |
| Terminal-Bench | Active | 56.4% | 80 | 2026-07-04 | independent | Source ↗ |
| OpenAI-MRCR | Active | 76.9% | 80 | 2026-07-03 | independent | Source ↗ |
| SimpleQA | Active | 47.6% | 90 | 2026-07-03 | independent | Source ↗ |
| CodeContests | Deprecated | 52.3% | 10 | 2026-07-03 | independent | Source ↗ |
| ComplexBench | Nearing Saturation | 52.5% | 80 | 2026-07-02 | independent | Source ↗ |
| SWE-bench Pro | Active | 49.9% | 90 | 2026-07-02 | independent | Source ↗ |
| LongBench-Pro | Active | 75% | 80 | 2026-07-02 | independent | Source ↗ |
| ZeroSCROLLS | Nearing Saturation | 73.2% | 78 | 2026-07-01 | independent | Source ↗ |
| MultiPL-E | Deprecated | 74.1% | 10 | 2026-07-01 | independent | Source ↗ |
| LiveCodeBench | Saturated | 66% | 39 | 2026-06-30 | independent | Source ↗ |
| AgentIF | Nearing Saturation | 43.5% | 80 | 2026-06-30 | independent | Source ↗ |
| MATH | Saturated | 89.2% | 25 | 2026-06-30 | independent | Source ↗ |
| InfiniteBench | Nearing Saturation | 77.8% | 65 | 2026-06-30 | independent | Source ↗ |
| Mercury | Deprecated | 70.8% | 10 | 2026-06-29 | independent | Source ↗ |
| SCROLLS | Nearing Saturation | 73.2% | 65 | 2026-06-29 | independent | Source ↗ |
| MMLU-Pro | Nearing Saturation | 84.2% | 75 | 2026-06-29 | independent | Source ↗ |
| RULER | Nearing Saturation | 84.2% | 84 | 2026-06-28 | independent | Source ↗ |
| TabMWP | Saturated | 83.7% | 25 | 2026-06-28 | independent | Source ↗ |
| CyberSecEval | Deprecated | 72.5% | 10 | 2026-06-28 | independent | Source ↗ |
| SWE-bench Verified | Saturated | 62.8% | 25 | 2026-06-27 | independent | Source ↗ |
| MMLU-Redux | Saturated | 89.7% | 25 | 2026-06-27 | independent | Source ↗ |
| QuALITY | Nearing Saturation | 80.5% | 78 | 2026-06-27 | independent | Source ↗ |
| LongBench | Nearing Saturation | 78.7% | 65 | 2026-06-26 | independent | Source ↗ |
| SWE-bench | Deprecated | 54.7% | 10 | 2026-06-26 | independent | Source ↗ |
| MGSM | Nearing Saturation | 86.5% | 65 | 2026-06-26 | independent | Source ↗ |
| AGIEval | Saturated | 78.7% | 25 | 2026-06-25 | independent | Source ↗ |
| QASPER | Nearing Saturation | 75.9% | 65 | 2026-06-25 | independent | Source ↗ |
| TruthfulQA | Saturated | 80.5% | 25 | 2026-06-25 | independent | Source ↗ |
| GSM-Hard | Deprecated | 87.4% | 10 | 2026-06-24 | independent | Source ↗ |
| SuperGLUE | Saturated | 93.3% | 25 | 2026-06-24 | independent | Source ↗ |
| CodeXGLUE | Deprecated | 75.7% | 23 | 2026-06-23 | independent | Source ↗ |
| MT-Bench | Nearing Saturation | 6.6% | 75 | 2026-06-23 | independent | Source ↗ |
| BIG-Bench Hard | Saturated | 89.7% | 25 | 2026-06-23 | independent | Source ↗ |
| WinoGrande | Saturated | 95.2% | 28 | 2026-06-22 | independent | Source ↗ |
| MBPP+ | Nearing Saturation | 82.1% | 65 | 2026-06-22 | independent | Source ↗ |
| IFEval | Deprecated | 62.7% | 25 | 2026-06-22 | independent | Source ↗ |
| GPQA | Nearing Saturation | 80.5% | 78 | 2026-06-22 | independent | Source ↗ |
| Needle-in-a-Haystack | Saturated | 98.3% | 44 | 2026-06-21 | independent | Source ↗ |
| HumanEval+ | Saturated | 86.9% | 25 | 2026-06-21 | independent | Source ↗ |
| CommonsenseQA | Saturated | 96.1% | 28 | 2026-06-21 | independent | Source ↗ |
| MBPP | Saturated | 85.3% | 25 | 2026-06-20 | independent | Source ↗ |
| GSM8K | Saturated | 98.3% | 25 | 2026-06-20 | independent | Source ↗ |
| GLUE | Saturated | 98.8% | 28 | 2026-06-20 | independent | Source ↗ |
| MMLU | Saturated | 93.3% | 25 | 2026-06-19 | independent | Source ↗ |
| HumanEval | Saturated | 91.8% | 25 | 2026-06-19 | independent | Source ↗ |
| ARC-AGI | Saturated | 86% | 25 | 2026-06-19 | independent | Source ↗ |
| LiveBench (Continuous) | Active | 77.4% | 94 | 2025-02-01 | independent | Source ↗ |
Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run qwen3-235b
# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "qwen3-235b",
"messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
}'Evaluation & Sourcing Notes
Scores listed for Qwen3-235B 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.