Model comparison
o3 vs Qwen2.5 72B Instruct
o3 is the stronger model overall, scoring 47.5 to 31.9 on the Noometry Index.
Last verified . 32 shared benchmarks.
Alibaba (Qwen)Summary
- They share 32 benchmarks with published results for both. o3 scores higher in 9 categories and Qwen2.5 72B Instruct in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where o3 leads 50.2 to 19.3.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 84.4% for o3 and 8.1% for Qwen2.5 72B Instruct.
- Qwen2.5 72B Instruct is cheaper at $1.40 / $5.60 per million input/output tokens, against $2 / $8 for o3.
- o3 accepts more context: 200K tokens versus 131K.
- Qwen2.5 72B Instruct has downloadable open weights; the other is API-only.
Side by side
| o3 | Qwen2.5 72B Instruct | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 47.5 | 31.9 |
| Released | 2025-04-16 | 2024-09 |
| Weights | Proprietary | Open |
| Context window | 200K | 131K |
| Max output | 100K | 8K |
| Input $ / M tokens | $2 | $1.40 |
| Output $ / M tokens | $8 | $5.60 |
| Results tracked | 63 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o3 leads
o3: 46.8 (#64), Qwen2.5 72B Instruct: 33.2 (#260)
| Benchmark | o3 | Qwen2.5 72B Instruct |
|---|---|---|
| WeirdML | 52.4% | 16% |
| LMArena Coding | 1408 | 1292 |
| SWE-bench Verified | 62.3% | — |
| SWE-bench Verified (bash only) | 58.4% | — |
| Aider Polyglot | 81.3% | — |
| GSO | 8.8% | — |
| BigCodeBench Instruct | — | 45.8% |
| BigCodeBench Complete | — | 55.9% |
| CadEval | 74% | — |
| ALE-Bench | 933.55 | — |
Agentic & Tool Use o3 leads
o3: 34.5 (#44), Qwen2.5 72B Instruct: 22.1 (#133)
| Benchmark | o3 | Qwen2.5 72B Instruct |
|---|---|---|
| METR Time Horizons | 65.4% | 35.8% |
| Berkeley Function Calling Leaderboard | 63% | — |
| GDPval | 30.8% | — |
| TheAgentCompany | — | 5.7% |
| DeepResearch Bench | 45.2% | — |
| OSWorld | 23% | — |
| BALROG | — | 16.2% |
| LMArena Search | 1144 | — |
Reasoning o3 leads
o3: 32.0 (#78), Qwen2.5 72B Instruct: 22.3 (#199)
| Benchmark | o3 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1402 | 1271 |
| DTBench | 84.8% | 62.9% |
| LMCA | 39.7% | 13.4% |
| Epoch Capabilities Index | 146.86 | 129 |
| ForecastBench | 62.5 | 57.5 |
| ARC-AGI-2 | 6.5% | — |
| SimpleBench | 53.1% | — |
| Kagi LLM Benchmark | 67.6% | — |
| ARC-AGI-1 | 60.8% | — |
| CritPt | 1.4% | — |
| Chess Puzzles | 38% | — |
| EnigmaEval | 13.1% | — |
| Mystery Game Puzzles | 29% | — |
| BIG-Bench Hard | — | 79.8% |
| HellaSwag | — | 84.8% |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |
Math o3 leads
o3: 50.2 (#58), Qwen2.5 72B Instruct: 19.3 (#287)
| Benchmark | o3 | Qwen2.5 72B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 84.4% | 8.1% |
| Omni-MATH | 71.4% | 33% |
| LMArena Math | 1426 | 1283 |
| MATH Level 5 | 97.8% | 63.2% |
| FrontierMath (Tiers 1-3) | 33.3% | — |
| FrontierMath (Feb 2025 set) | 18.7% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge o3 leads
o3: 54.6 (#52), Qwen2.5 72B Instruct: 27.0 (#253)
| Benchmark | o3 | Qwen2.5 72B Instruct |
|---|---|---|
| GPQA Diamond | 81.8% | 49.1% |
| MMLU-Pro | 85.9% | 63.1% |
| Confabulations | 14.4% | 19.1% |
| GPQA (HELM) | 75.3% | 42.6% |
| LMArena Expert | 1402 | 1245 |
| Humanity's Last Exam | 20.3% | — |
| SimpleQA Verified | 49.4% | — |
| ARC (AI2) Challenge | — | 94.5% |
| MMLU | — | 85.3% |
| TriviaQA | — | 71.9% |
Multimodal Not comparable
o3: 41.4 (#36), Qwen2.5 72B Instruct: —
| Benchmark | o3 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Vision | 1214 | — |
| GeoBench | 74% | — |
| VPCT | 52% | — |
Multilingual o3 leads
o3: 51.7 (#105), Qwen2.5 72B Instruct: 41.0 (#213)
| Benchmark | o3 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | 1401 | 1252 |
| LMArena Chinese | 1437 | 1272 |
| LMArena French | 1430 | 1280 |
| LMArena German | 1420 | 1234 |
| LMArena Japanese | 1403 | 1180 |
| LMArena Korean | 1370 | 1188 |
| LMArena Russian | 1406 | 1264 |
| LMArena Spanish | 1395 | 1256 |
Instruction Following o3 leads
o3: 72.8 (#127), Qwen2.5 72B Instruct: 65.5 (#221)
| Benchmark | o3 | Qwen2.5 72B Instruct |
|---|---|---|
| IFEval | 86.9% | 80.6% |
| LMArena Instruction Following | 1368 | 1254 |
Long Context o3 leads
o3: 53.3 (#6), Qwen2.5 72B Instruct: 38.9 (#188)
| Benchmark | o3 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | 1372 | 1282 |
| Fiction.LiveBench | 88.9% | — |
| CL-bench | 17.8% | — |
Writing & Preference o3 leads
o3: 63.5 (#64), Qwen2.5 72B Instruct: 46.7 (#215)
| Benchmark | o3 | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | 1410 | 1269 |
| LMArena Creative Writing | 1359 | 1221 |
| WildBench | 86.1% | 80.2% |
| LMArena Multi-Turn | 1405 | 1272 |
| Short-Story Creative Writing | 83.9% | — |
| EQ-Bench Creative Writing | 1676 | — |
Frequently asked questions
Is o3 better than Qwen2.5 72B Instruct?
o3 is the stronger model overall, scoring 47.5 to 31.9 on the Noometry Index.
Which is cheaper, o3 or Qwen2.5 72B Instruct?
Qwen2.5 72B Instruct is cheaper. It lists at $1.40 per million input tokens and $5.60 per million output tokens; o3 lists at $2 and $8.
Is o3 or Qwen2.5 72B Instruct better for coding?
o3 scores higher on coding benchmarks: 46.8 versus 33.2 in the Noometry coding category.
Which has the bigger context window?
o3 does, with 200K tokens against 131K.
How many benchmarks do o3 and Qwen2.5 72B Instruct share?
32 benchmarks have published results for both models. o3 has 63 scored results on Noometry and Qwen2.5 72B Instruct has 43.