Model comparison
o3 vs Qwen3 235B-A22B
o3 is the stronger model overall, scoring 47.5 to 43.5 on the Noometry Index. Qwen3 235B-A22B costs 2.9× less per token, which makes it the better buy when o3's lead doesn't matter for your workload.
Last verified . 46 shared benchmarks.
Alibaba (Qwen)Summary
- They share 46 benchmarks with published results for both. o3 scores higher in 7 categories and Qwen3 235B-A22B in 2 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where o3 leads 32.0 to 15.7.
- The biggest single-benchmark swing is ARC-AGI-1: 60.8% for o3 and 11% for Qwen3 235B-A22B.
- Qwen3 235B-A22B is cheaper at $0.70 / $2.80 per million input/output tokens, against $2 / $8 for o3.
- o3 accepts more context: 200K tokens versus 131K.
- Qwen3 235B-A22B has downloadable open weights; the other is API-only.
Side by side
| o3 | Qwen3 235B-A22B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 47.5 | 43.5 |
| Released | 2025-04-16 | 2025-04 |
| Weights | Proprietary | Open |
| Context window | 200K | 131K |
| Max output | 100K | 16K |
| Input $ / M tokens | $2 | $0.70 |
| Output $ / M tokens | $8 | $2.80 |
| Results tracked | 63 | 49 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o3 leads
o3: 46.8 (#64), Qwen3 235B-A22B: 44.3 (#75)
| Benchmark | o3 | Qwen3 235B-A22B |
|---|---|---|
| Aider Polyglot | 81.3% | 59.6% |
| WeirdML | 52.4% | 41% |
| LMArena Coding | 1408 | 1445 |
| SWE-bench Verified | 62.3% | — |
| SWE-bench Verified (bash only) | 58.4% | — |
| SciCode | — | 42.4% |
| GSO | 8.8% | — |
| CadEval | 74% | — |
| ALE-Bench | 933.55 | — |
Agentic & Tool Use Too close to call
o3: 34.5 (#44), Qwen3 235B-A22B: 33.9 (#51)
| Benchmark | o3 | Qwen3 235B-A22B |
|---|---|---|
| Berkeley Function Calling Leaderboard | 63% | 52.1% |
| GDPval | 30.8% | — |
| DeepResearch Bench | 45.2% | — |
| OSWorld | 23% | — |
| LMArena Search | 1144 | — |
| METR Time Horizons | 65.4% | — |
| Vending-Bench 2 | — | -11.34 |
Reasoning o3 leads
o3: 32.0 (#78), Qwen3 235B-A22B: 15.7 (#311)
| Benchmark | o3 | Qwen3 235B-A22B |
|---|---|---|
| ARC-AGI-2 | 6.5% | 1.3% |
| SimpleBench | 53.1% | 31% |
| Kagi LLM Benchmark | 67.6% | 69.4% |
| ARC-AGI-1 | 60.8% | 11% |
| CritPt | 1.4% | 0% |
| Chess Puzzles | 38% | 12% |
| LMArena Hard Prompts | 1402 | 1433 |
| Mystery Game Puzzles | 29% | 9% |
| DTBench | 84.8% | 80.3% |
| LMCA | 39.7% | 29.3% |
| Epoch Capabilities Index | 146.86 | 143.85 |
| ForecastBench | 62.5 | 59.7 |
| EnigmaEval | 13.1% | — |
Math Too close to call
o3: 50.2 (#58), Qwen3 235B-A22B: 50.4 (#57)
| Benchmark | o3 | Qwen3 235B-A22B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 84.4% | 86.7% |
| Omni-MATH | 71.4% | 71.8% |
| LMArena Math | 1426 | 1432 |
| MATH Level 5 | 97.8% | 68.9% |
| FrontierMath (Feb 2025 set) | 18.7% | 8.5% |
| FrontierMath Tier 4 (v1) | 2.1% | 0% |
| FrontierMath (Tiers 1-3) | 33.3% | — |
Knowledge o3 leads
o3: 54.6 (#52), Qwen3 235B-A22B: 49.6 (#73)
| Benchmark | o3 | Qwen3 235B-A22B |
|---|---|---|
| GPQA Diamond | 81.8% | 80.1% |
| SimpleQA Verified | 49.4% | 40.4% |
| MMLU-Pro | 85.9% | 84.4% |
| Confabulations | 14.4% | 15.6% |
| GPQA (HELM) | 75.3% | 72.7% |
| LMArena Expert | 1402 | 1463 |
| Humanity's Last Exam | 20.3% | — |
| Vectara Hallucination Rate | — | 9.3% |
Multimodal Not comparable
o3: 41.4 (#36), Qwen3 235B-A22B: —
| Benchmark | o3 | Qwen3 235B-A22B |
|---|---|---|
| LMArena Vision | 1214 | — |
| GeoBench | 74% | — |
| VPCT | 52% | — |
Multilingual Too close to call
o3: 51.7 (#105), Qwen3 235B-A22B: 52.3 (#89)
| Benchmark | o3 | Qwen3 235B-A22B |
|---|---|---|
| LMArena Non-English | 1401 | 1409 |
| LMArena Chinese | 1437 | 1481 |
| LMArena French | 1430 | 1445 |
| LMArena German | 1420 | 1433 |
| LMArena Japanese | 1403 | 1399 |
| LMArena Korean | 1370 | 1391 |
| LMArena Russian | 1406 | 1411 |
| LMArena Spanish | 1395 | 1430 |
Instruction Following Too close to call
o3: 72.8 (#127), Qwen3 235B-A22B: 72.6 (#136)
| Benchmark | o3 | Qwen3 235B-A22B |
|---|---|---|
| IFEval | 86.9% | 83.5% |
| LMArena Instruction Following | 1368 | 1408 |
Long Context o3 leads
o3: 53.3 (#6), Qwen3 235B-A22B: 46.1 (#26)
| Benchmark | o3 | Qwen3 235B-A22B |
|---|---|---|
| Fiction.LiveBench | 88.9% | 75% |
| LMArena Longer Query | 1372 | 1426 |
| CL-bench | 17.8% | — |
Writing & Preference o3 leads
o3: 63.5 (#64), Qwen3 235B-A22B: 59.6 (#108)
| Benchmark | o3 | Qwen3 235B-A22B |
|---|---|---|
| LMArena Text | 1410 | 1419 |
| LMArena Creative Writing | 1359 | 1384 |
| Short-Story Creative Writing | 83.9% | 83% |
| EQ-Bench Creative Writing | 1676 | 1366 |
| WildBench | 86.1% | 86.6% |
| LMArena Multi-Turn | 1405 | 1432 |
Frequently asked questions
Is o3 better than Qwen3 235B-A22B?
o3 is the stronger model overall, scoring 47.5 to 43.5 on the Noometry Index. Qwen3 235B-A22B costs 2.9× less per token, which makes it the better buy when o3's lead doesn't matter for your workload.
Which is cheaper, o3 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; o3 lists at $2 and $8.
Is o3 or Qwen3 235B-A22B better for coding?
o3 scores higher on coding benchmarks: 46.8 versus 44.3 in the Noometry coding category.
Which has the bigger context window?
o3 does, with 200K tokens against 131K.
How many benchmarks do o3 and Qwen3 235B-A22B share?
46 benchmarks have published results for both models. o3 has 63 scored results on Noometry and Qwen3 235B-A22B has 49.