Model comparison
GPT-5.2 vs o3
GPT-5.2 is the stronger model overall, scoring 54.1 to 47.5 on the Noometry Index.
Last verified . 49 shared benchmarks.
Summary
- They share 49 benchmarks with published results for both. GPT-5.2 scores higher in 9 categories and o3 in 1 category; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.2 leads 50.2 to 32.0.
- The biggest single-benchmark swing is ARC-AGI-2: 52.9% for GPT-5.2 and 6.5% for o3.
- o3 is cheaper at $2 / $8 per million input/output tokens, against $1.75 / $14 for GPT-5.2.
- GPT-5.2 accepts more context: 400K tokens versus 200K.
Side by side
| GPT-5.2 | o3 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 54.1 | 47.5 |
| Released | 2025-12-11 | 2025-04-16 |
| Weights | Proprietary | Proprietary |
| Context window | 400K | 200K |
| Max output | 128K | 100K |
| Input $ / M tokens | $1.75 | $2 |
| Output $ / M tokens | $14 | $8 |
| Results tracked | 67 | 63 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.2 leads
GPT-5.2: 51.6 (#37), o3: 46.8 (#64)
| Benchmark | GPT-5.2 | o3 |
|---|---|---|
| SWE-bench Verified | 73.8% | 62.3% |
| SWE-bench Verified (bash only) | 72.8% | 58.4% |
| GSO | 27.4% | 8.8% |
| WeirdML | 72.2% | 52.4% |
| LMArena Coding | 1447 | 1408 |
| ALE-Bench | 1,294 | 933.55 |
| Aider Polyglot | — | 81.3% |
| LMArena WebDev | 1416 | — |
| SWE-bench Multilingual | 66.7% | — |
| CadEval | — | 74% |
| AlgoTune | 2.05 | — |
Agentic & Tool Use GPT-5.2 leads
GPT-5.2: 40.2 (#24), o3: 34.5 (#44)
| Benchmark | GPT-5.2 | o3 |
|---|---|---|
| Berkeley Function Calling Leaderboard | 55.9% | 63% |
| GDPval | 49.7% | 30.8% |
| DeepResearch Bench | 41.1% | 45.2% |
| LMArena Search | 1207 | 1144 |
| METR Time Horizons | 75.3% | 65.4% |
| Terminal-Bench | 64.9% | — |
| Remote Labor Index | 2.5% | — |
| τ²-bench Airline | 83% | — |
| τ²-bench Banking | 32.2% | — |
| τ²-bench Retail | 81.6% | — |
| τ²-bench Telecom | 89.7% | — |
| OSWorld | — | 23% |
| Vending-Bench 2 | 3,591 | — |
Reasoning GPT-5.2 leads
GPT-5.2: 50.2 (#35), o3: 32.0 (#78)
| Benchmark | GPT-5.2 | o3 |
|---|---|---|
| ARC-AGI-2 | 52.9% | 6.5% |
| SimpleBench | 45.8% | 53.1% |
| Kagi LLM Benchmark | 73.3% | 67.6% |
| ARC-AGI-1 | 86.2% | 60.8% |
| Chess Puzzles | 49% | 38% |
| EnigmaEval | 10.4% | 13.1% |
| LMArena Hard Prompts | 1445 | 1402 |
| Mystery Game Puzzles | 23% | 29% |
| DTBench | 90.9% | 84.8% |
| LMCA | 43.9% | 39.7% |
| Epoch Capabilities Index | 153.45 | 146.86 |
| ForecastBench | 60.1 | 62.5 |
| NYT Connections (extended) | 83.6% | — |
| CritPt | — | 1.4% |
| EBR-Bench | 23% | — |
Math GPT-5.2 leads
GPT-5.2: 60.0 (#38), o3: 50.2 (#58)
| Benchmark | GPT-5.2 | o3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 67.4% | 33.3% |
| OTIS Mock AIME 2024-2025 | 96.1% | 84.4% |
| LMArena Math | 1440 | 1426 |
| FrontierMath (Feb 2025 set) | 40.7% | 18.7% |
| FrontierMath Tier 4 (v1) | 18.8% | 2.1% |
| FrontierMath Tier 4 | 31.7% | — |
| MathArena Final-Answer Competitions | 72% | — |
| ProofBench | 15% | — |
| Omni-MATH | — | 71.4% |
| MATH Level 5 | — | 97.8% |
Knowledge GPT-5.2 leads
GPT-5.2: 59.3 (#32), o3: 54.6 (#52)
| Benchmark | GPT-5.2 | o3 |
|---|---|---|
| GPQA Diamond | 91.4% | 81.8% |
| Humanity's Last Exam | 27.8% | 20.3% |
| SimpleQA Verified | 37.1% | 49.4% |
| LMArena Expert | 1445 | 1402 |
| MMLU-Pro | — | 85.9% |
| Confabulations | — | 14.4% |
| Vectara Hallucination Rate | 8.4% | — |
| GPQA (HELM) | — | 75.3% |
Multimodal GPT-5.2 leads
GPT-5.2: 51.3 (#7), o3: 41.4 (#36)
| Benchmark | GPT-5.2 | o3 |
|---|---|---|
| LMArena Vision | 1268 | 1214 |
| VPCT | 84% | 52% |
| GeoBench | — | 74% |
| Furniture Assembly | 38.3% | — |
| LMArena Document | 1405 | — |
Multilingual GPT-5.2 leads
GPT-5.2: 53.4 (#67), o3: 51.7 (#105)
| Benchmark | GPT-5.2 | o3 |
|---|---|---|
| LMArena Non-English | 1425 | 1401 |
| LMArena Chinese | 1460 | 1437 |
| LMArena French | 1455 | 1430 |
| LMArena German | 1448 | 1420 |
| LMArena Japanese | 1420 | 1403 |
| LMArena Korean | 1392 | 1370 |
| LMArena Russian | 1440 | 1406 |
| LMArena Spanish | 1433 | 1395 |
Instruction Following GPT-5.2 leads
GPT-5.2: 74.7 (#89), o3: 72.8 (#127)
| Benchmark | GPT-5.2 | o3 |
|---|---|---|
| LMArena Instruction Following | 1417 | 1368 |
| IFEval | — | 86.9% |
Long Context o3 leads
GPT-5.2: 44.0 (#78), o3: 53.3 (#6)
| Benchmark | GPT-5.2 | o3 |
|---|---|---|
| CL-bench | 18.2% | 17.8% |
| LMArena Longer Query | 1428 | 1372 |
| Fiction.LiveBench | — | 88.9% |
Writing & Preference GPT-5.2 leads
GPT-5.2: 66.8 (#32), o3: 63.5 (#64)
| Benchmark | GPT-5.2 | o3 |
|---|---|---|
| LMArena Text | 1439 | 1410 |
| LMArena Creative Writing | 1401 | 1359 |
| EQ-Bench Creative Writing | 1703 | 1676 |
| LMArena Multi-Turn | 1458 | 1405 |
| Short-Story Creative Writing | — | 83.9% |
| WildBench | — | 86.1% |
Frequently asked questions
Is GPT-5.2 better than o3?
GPT-5.2 is the stronger model overall, scoring 54.1 to 47.5 on the Noometry Index.
Which is cheaper, GPT-5.2 or o3?
o3 is cheaper. It lists at $2 per million input tokens and $8 per million output tokens; GPT-5.2 lists at $1.75 and $14.
Is GPT-5.2 or o3 better for coding?
GPT-5.2 scores higher on coding benchmarks: 51.6 versus 46.8 in the Noometry coding category.
Which has the bigger context window?
GPT-5.2 does, with 400K tokens against 200K.
How many benchmarks do GPT-5.2 and o3 share?
49 benchmarks have published results for both models. GPT-5.2 has 67 scored results on Noometry and o3 has 63.