Model comparison
GPT-4o vs Qwen3 235B-A22B
Qwen3 235B-A22B is the stronger model overall, scoring 43.5 to 28.6 on the Noometry Index.
Last verified . 42 shared benchmarks.
Alibaba (Qwen)Summary
- They share 42 benchmarks with published results for both. GPT-4o scores higher in 0 categories and Qwen3 235B-A22B in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen3 235B-A22B leads 50.4 to 10.6.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 6.4% for GPT-4o and 86.7% for Qwen3 235B-A22B.
- Qwen3 235B-A22B is cheaper at $0.70 / $2.80 per million input/output tokens, against $2.50 / $10 for GPT-4o.
- Qwen3 235B-A22B accepts more context: 131K tokens versus 128K.
- Qwen3 235B-A22B has downloadable open weights; the other is API-only.
Side by side
| GPT-4o | Qwen3 235B-A22B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 28.6 | 43.5 |
| Released | 2024-05-13 | 2025-04 |
| Weights | Proprietary | Open |
| Context window | 128K | 131K |
| Max output | 16K | 16K |
| Input $ / M tokens | $2.50 | $0.70 |
| Output $ / M tokens | $10 | $2.80 |
| Results tracked | 72 | 49 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3 235B-A22B leads
GPT-4o: 24.8 (#328), Qwen3 235B-A22B: 44.3 (#75)
| Benchmark | GPT-4o | Qwen3 235B-A22B |
|---|---|---|
| Aider Polyglot | 45.3% | 59.6% |
| WeirdML | 25.1% | 41% |
| LMArena Coding | 1297 | 1445 |
| SWE-bench Verified | 31% | — |
| SWE-bench Verified (bash only) | 21.6% | — |
| SciCode | — | 42.4% |
| GSO | 0% | — |
| BigCodeBench Instruct | 51.1% | — |
| LiveBench Coding | 51.4% | — |
| BigCodeBench Complete | 61.1% | — |
| CadEval | 26% | — |
| HumanEval+ | 87.2% | — |
| MBPP+ | 72.2% | — |
Agentic & Tool Use Qwen3 235B-A22B leads
GPT-4o: 21.0 (#141), Qwen3 235B-A22B: 33.9 (#51)
| Benchmark | GPT-4o | Qwen3 235B-A22B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 52.1% |
| GDPval | 9.9% | — |
| TheAgentCompany | 8.6% | — |
| Cybench | 12.5% | — |
| BALROG | 32.3% | — |
| LMArena Search | 1006 | — |
| METR Time Horizons | 40.8% | — |
| Vending-Bench 2 | — | -11.34 |
Reasoning Qwen3 235B-A22B leads
GPT-4o: 9.4 (#343), Qwen3 235B-A22B: 15.7 (#311)
| Benchmark | GPT-4o | Qwen3 235B-A22B |
|---|---|---|
| ARC-AGI-2 | 0% | 1.3% |
| SimpleBench | 17.8% | 31% |
| ARC-AGI-1 | 4.5% | 11% |
| CritPt | 0% | 0% |
| Chess Puzzles | 13% | 12% |
| LMArena Hard Prompts | 1281 | 1433 |
| DTBench | 64.5% | 80.3% |
| LMCA | 16.6% | 29.3% |
| Epoch Capabilities Index | 128.97 | 143.85 |
| ForecastBench | 57.7 | 59.7 |
| Kagi LLM Benchmark | — | 69.4% |
| EnigmaEval | 0.8% | — |
| LiveBench Reasoning | 55.8% | — |
| Mystery Game Puzzles | — | 9% |
| LiveBench Data Analysis | 60.9% | — |
| LiveBench | 55.3% | — |
Math Qwen3 235B-A22B leads
GPT-4o: 10.6 (#312), Qwen3 235B-A22B: 50.4 (#57)
| Benchmark | GPT-4o | Qwen3 235B-A22B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 6.4% | 86.7% |
| Omni-MATH | 29.3% | 71.8% |
| LMArena Math | 1285 | 1432 |
| MATH Level 5 | 53.3% | 68.9% |
| FrontierMath (Feb 2025 set) | 0.3% | 8.5% |
| FrontierMath (Tiers 1-3) | 0.4% | — |
| LiveBench Math | 49.5% | — |
| FrontierMath Tier 4 (v1) | — | 0% |
Knowledge Qwen3 235B-A22B leads
GPT-4o: 28.8 (#242), Qwen3 235B-A22B: 49.6 (#73)
| Benchmark | GPT-4o | Qwen3 235B-A22B |
|---|---|---|
| GPQA Diamond | 49.2% | 80.1% |
| SimpleQA Verified | 26% | 40.4% |
| MMLU-Pro | 71.3% | 84.4% |
| Confabulations | 15.3% | 15.6% |
| Vectara Hallucination Rate | 9.6% | 9.3% |
| GPQA (HELM) | 52% | 72.7% |
| LMArena Expert | 1250 | 1463 |
| Humanity's Last Exam | 2.7% | — |
| MMLU | 88.1% | — |
Multimodal Not comparable
GPT-4o: 34.5 (#91), Qwen3 235B-A22B: —
| Benchmark | GPT-4o | Qwen3 235B-A22B |
|---|---|---|
| LMArena Vision | 1137 | — |
| Video-MME | 71.9% | — |
| GeoBench | 71% | — |
| VPCT | 40% | — |
| ScienceQA | 88.5% | — |
Multilingual Qwen3 235B-A22B leads
GPT-4o: 43.2 (#186), Qwen3 235B-A22B: 52.3 (#89)
| Benchmark | GPT-4o | Qwen3 235B-A22B |
|---|---|---|
| LMArena Non-English | 1283 | 1409 |
| LMArena Chinese | 1277 | 1481 |
| LMArena French | 1304 | 1445 |
| LMArena German | 1282 | 1433 |
| LMArena Japanese | 1257 | 1399 |
| LMArena Korean | 1234 | 1391 |
| LMArena Russian | 1286 | 1411 |
| LMArena Spanish | 1292 | 1430 |
Instruction Following Qwen3 235B-A22B leads
GPT-4o: 66.6 (#207), Qwen3 235B-A22B: 72.6 (#136)
| Benchmark | GPT-4o | Qwen3 235B-A22B |
|---|---|---|
| IFEval | 81.7% | 83.5% |
| LMArena Instruction Following | 1278 | 1408 |
| LiveBench Instruction Following | 68.6% | — |
Long Context Qwen3 235B-A22B leads
GPT-4o: 39.4 (#179), Qwen3 235B-A22B: 46.1 (#26)
| Benchmark | GPT-4o | Qwen3 235B-A22B |
|---|---|---|
| Fiction.LiveBench | 66.7% | 75% |
| LMArena Longer Query | 1289 | 1426 |
Writing & Preference Qwen3 235B-A22B leads
GPT-4o: 52.6 (#166), Qwen3 235B-A22B: 59.6 (#108)
| Benchmark | GPT-4o | Qwen3 235B-A22B |
|---|---|---|
| LMArena Text | 1300 | 1419 |
| LMArena Creative Writing | 1292 | 1384 |
| Short-Story Creative Writing | 81.8% | 83% |
| WildBench | 82.8% | 86.6% |
| LMArena Multi-Turn | 1302 | 1432 |
| EQ-Bench Creative Writing | — | 1366 |
| LiveBench Language | 47.6% | — |
Frequently asked questions
Is GPT-4o better than Qwen3 235B-A22B?
Qwen3 235B-A22B is the stronger model overall, scoring 43.5 to 28.6 on the Noometry Index.
Which is cheaper, GPT-4o or Qwen3 235B-A22B?
Qwen3 235B-A22B is cheaper. It lists at $0.70 per million input tokens and $2.80 per million output tokens; GPT-4o lists at $2.50 and $10.
Is GPT-4o or Qwen3 235B-A22B better for coding?
Qwen3 235B-A22B scores higher on coding benchmarks: 44.3 versus 24.8 in the Noometry coding category.
Which has the bigger context window?
Qwen3 235B-A22B does, with 131K tokens against 128K.
How many benchmarks do GPT-4o and Qwen3 235B-A22B share?
42 benchmarks have published results for both models. GPT-4o has 72 scored results on Noometry and Qwen3 235B-A22B has 49.