Model comparison
o3 vs Qwen3.8 27B
o3 is the stronger model overall, scoring 47.5 to 46.0 on the Noometry Index. Qwen3.8 27B costs 3.1× less per token, which makes it the better buy when o3's lead doesn't matter for your workload.
Last verified . 25 shared benchmarks.
Alibaba (Qwen)Summary
- They share 25 benchmarks with published results for both. o3 scores higher in 5 categories and Qwen3.8 27B in 5 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where o3 leads 50.2 to 37.1.
- The biggest single-benchmark swing is ARC-AGI-2: 6.5% for o3 and 42.4% for Qwen3.8 27B.
- Qwen3.8 27B is cheaper at $0.99 / $1.49 per million input/output tokens, against $2 / $8 for o3.
- Qwen3.8 27B accepts more context: 262K tokens versus 200K.
- Qwen3.8 27B has downloadable open weights; the other is API-only.
Side by side
| o3 | Qwen3.8 27B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 47.5 | 46.0 |
| Released | 2025-04-16 | 2026-08-14 |
| Weights | Proprietary | Open |
| Context window | 200K | 262K |
| Max output | 100K | 33K |
| Input $ / M tokens | $2 | $0.99 |
| Output $ / M tokens | $8 | $1.49 |
| Results tracked | 63 | 31 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 27B leads
o3: 46.8 (#64), Qwen3.8 27B: 50.5 (#44)
| Benchmark | o3 | Qwen3.8 27B |
|---|---|---|
| LMArena Coding | 1408 | 1482 |
| SWE-bench Verified | 62.3% | — |
| SWE-bench Verified (bash only) | 58.4% | — |
| Aider Polyglot | 81.3% | — |
| LMArena WebDev | — | 1593 |
| SciCode | — | 46.6% |
| GSO | 8.8% | — |
| WeirdML | 52.4% | — |
| CadEval | 74% | — |
| ALE-Bench | 933.55 | — |
Agentic & Tool Use o3 leads
o3: 34.5 (#44), Qwen3.8 27B: 32.9 (#57)
| Benchmark | o3 | Qwen3.8 27B |
|---|---|---|
| APEX-Agents | — | 47.5% |
| Berkeley Function Calling Leaderboard | 63% | — |
| GDPval | 30.8% | — |
| DeepResearch Bench | 45.2% | — |
| OSWorld | 23% | — |
| LMArena Search | 1144 | — |
| METR Time Horizons | 65.4% | — |
Reasoning Qwen3.8 27B leads
o3: 32.0 (#78), Qwen3.8 27B: 41.0 (#54)
| Benchmark | o3 | Qwen3.8 27B |
|---|---|---|
| ARC-AGI-2 | 6.5% | 42.4% |
| ARC-AGI-1 | 60.8% | 87.5% |
| CritPt | 1.4% | 5.4% |
| LMArena Hard Prompts | 1402 | 1460 |
| DTBench | 84.8% | 88% |
| LMCA | 39.7% | 41.4% |
| Epoch Capabilities Index | 146.86 | 149.38 |
| SimpleBench | 53.1% | — |
| Kagi LLM Benchmark | 67.6% | — |
| NYT Connections (extended) | — | 54.5% |
| Chess Puzzles | 38% | — |
| EnigmaEval | 13.1% | — |
| Mystery Game Puzzles | 29% | — |
| Surface Evolver Bench | — | 45% |
| ForecastBench | 62.5 | — |
Math o3 leads
o3: 50.2 (#58), Qwen3.8 27B: 37.1 (#161)
| Benchmark | o3 | Qwen3.8 27B |
|---|---|---|
| LMArena Math | 1426 | 1456 |
| FrontierMath (Tiers 1-3) | 33.3% | — |
| OTIS Mock AIME 2024-2025 | 84.4% | — |
| ProofBench | — | 16% |
| Omni-MATH | 71.4% | — |
| MATH Level 5 | 97.8% | — |
| FrontierMath (Feb 2025 set) | 18.7% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge o3 leads
o3: 54.6 (#52), Qwen3.8 27B: 41.6 (#109)
| Benchmark | o3 | Qwen3.8 27B |
|---|---|---|
| LMArena Expert | 1402 | 1482 |
| GPQA Diamond | 81.8% | — |
| Humanity's Last Exam | 20.3% | — |
| SimpleQA Verified | 49.4% | — |
| MMLU-Pro | 85.9% | — |
| Confabulations | 14.4% | — |
| GPQA (HELM) | 75.3% | — |
Multimodal Too close to call
o3: 41.4 (#36), Qwen3.8 27B: 41.3 (#37)
| Benchmark | o3 | Qwen3.8 27B |
|---|---|---|
| LMArena Vision | 1214 | 1271 |
| GeoBench | 74% | — |
| VPCT | 52% | — |
Multilingual Qwen3.8 27B leads
o3: 51.7 (#105), Qwen3.8 27B: 53.7 (#60)
| Benchmark | o3 | Qwen3.8 27B |
|---|---|---|
| LMArena Non-English | 1401 | 1430 |
| LMArena Chinese | 1437 | 1504 |
| LMArena French | 1430 | 1465 |
| LMArena German | 1420 | 1438 |
| LMArena Japanese | 1403 | 1384 |
| LMArena Korean | 1370 | 1393 |
| LMArena Russian | 1406 | 1415 |
| LMArena Spanish | 1395 | 1448 |
Instruction Following Qwen3.8 27B leads
o3: 72.8 (#127), Qwen3.8 27B: 75.8 (#53)
| Benchmark | o3 | Qwen3.8 27B |
|---|---|---|
| LMArena Instruction Following | 1368 | 1439 |
| IFEval | 86.9% | — |
Long Context o3 leads
o3: 53.3 (#6), Qwen3.8 27B: 44.3 (#70)
| Benchmark | o3 | Qwen3.8 27B |
|---|---|---|
| LMArena Longer Query | 1372 | 1450 |
| Fiction.LiveBench | 88.9% | — |
| CL-bench | 17.8% | — |
Writing & Preference Qwen3.8 27B leads
o3: 63.5 (#64), Qwen3.8 27B: 65.8 (#43)
| Benchmark | o3 | Qwen3.8 27B |
|---|---|---|
| LMArena Text | 1410 | 1441 |
| LMArena Creative Writing | 1359 | 1384 |
| EQ-Bench Creative Writing | 1676 | 1671 |
| LMArena Multi-Turn | 1405 | 1441 |
| Short-Story Creative Writing | 83.9% | — |
| WildBench | 86.1% | — |
Frequently asked questions
Is o3 better than Qwen3.8 27B?
o3 is the stronger model overall, scoring 47.5 to 46.0 on the Noometry Index. Qwen3.8 27B costs 3.1× less per token, which makes it the better buy when o3's lead doesn't matter for your workload.
Which is cheaper, o3 or Qwen3.8 27B?
Qwen3.8 27B is cheaper. It lists at $0.99 per million input tokens and $1.49 per million output tokens; o3 lists at $2 and $8.
Is o3 or Qwen3.8 27B better for coding?
Qwen3.8 27B scores higher on coding benchmarks: 50.5 versus 46.8 in the Noometry coding category.
Which has the bigger context window?
Qwen3.8 27B does, with 262K tokens against 200K.
How many benchmarks do o3 and Qwen3.8 27B share?
25 benchmarks have published results for both models. o3 has 63 scored results on Noometry and Qwen3.8 27B has 31.