# GPT-5.1 vs o3

> GPT-5.1 is the stronger model overall, scoring 48.8 to 47.4 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gpt-5-1-vs-o3
- Last updated: 2026-10-10
- Shared benchmarks: 49

## Summary

- They share 49 benchmarks with published results for both. GPT-5.1 scores higher in 6 categories and o3 in 4 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in instruction following, where GPT-5.1 leads 84.0 to 73.0.
- The biggest single-benchmark swing is GPQA (HELM): 44.2% for GPT-5.1 and 75.3% for o3.
- Both cost about the same: $1.25 input and $10 output per million tokens.
- GPT-5.1 accepts more context: 400K tokens versus 200K.

## Snapshot

| | GPT-5.1 | o3 |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 48.8 | 47.4 |
| Rank | 47 | 53 |
| Context | 400K | 200K |
| Input $/M | $1.25 | $2 |
| Output $/M | $10 | $8 |
| Weights | Proprietary | Proprietary |

## Coding

- GPT-5.1: 46.9 (#59)
- o3: 47.3 (#57)

| Benchmark | GPT-5.1 | o3 |
|---|---|---|
| SWE-bench Verified | 68% | 62.3% |
| SWE-bench Verified (bash only) | 66% | 58.4% |
| GSO | 13.7% | 8.8% |
| WeirdML | 60.8% | 52.4% |
| LMArena Coding | 1454 | 1408 |
| ALE-Bench | 1,192 | 933.55 |
| Aider Polyglot | — | 81.3% |
| LMArena WebDev | 1395 | — |
| SciCode | 43.3% | — |
| LiveBench Coding | 72.5% | — |
| CadEval | — | 74% |

## Agentic & Tool Use

- GPT-5.1: 32.5 (#55)
- o3: 34.3 (#42)

| Benchmark | GPT-5.1 | o3 |
|---|---|---|
| DeepResearch Bench | 42.8% | 45.2% |
| LMArena Search | 1199 | 1144 |
| Terminal-Bench | 47.6% | — |
| Berkeley Function Calling Leaderboard | — | 63% |
| GDPval | — | 30.8% |
| OSWorld | — | 23% |
| METR Time Horizons | — | 65.4% |
| Vending-Bench 2 | 1,473 | — |

## Reasoning

- GPT-5.1: 39.9 (#50)
- o3: 32.1 (#66)

| Benchmark | GPT-5.1 | o3 |
|---|---|---|
| ARC-AGI-2 | 17.6% | 6.5% |
| SimpleBench | 53.2% | 53.1% |
| ARC-AGI-1 | 72.8% | 60.8% |
| CritPt | 4.9% | 1.4% |
| Chess Puzzles | 32% | 38% |
| EnigmaEval | 11.2% | 13.1% |
| LMArena Hard Prompts | 1457 | 1402 |
| Mystery Game Puzzles | 19% | 29% |
| DTBench | 90.1% | 84.8% |
| LMCA | 43.9% | 39.7% |
| Epoch Capabilities Index | 149.66 | 146.88 |
| ForecastBench | 58.1 | 62.5 |
| Kagi LLM Benchmark | — | 67.6% |
| LiveBench Reasoning | 95.8% | — |
| LiveBench Data Analysis | 72.1% | — |
| LiveBench | 78.8% | — |

## Math

- GPT-5.1: 50.3 (#45)
- o3: 48.9 (#49)

| Benchmark | GPT-5.1 | o3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.6% | 84.4% |
| Omni-MATH | 46.4% | 71.4% |
| LMArena Math | 1446 | 1426 |
| FrontierMath (Feb 2025 set) | 31% | 18.7% |
| FrontierMath Tier 4 (v1) | 12.5% | 2.1% |
| FrontierMath (Tiers 1-3) | — | 33.3% |
| LiveBench Math | 94.5% | — |
| MATH Level 5 | — | 97.8% |

## Knowledge

- GPT-5.1: 50.5 (#65)
- o3: 54.5 (#47)

| Benchmark | GPT-5.1 | o3 |
|---|---|---|
| GPQA Diamond | 87.6% | 81.8% |
| Humanity's Last Exam | 23.7% | 20.3% |
| SimpleQA Verified | 48% | 49.4% |
| MMLU-Pro | 57.9% | 85.9% |
| GPQA (HELM) | 44.2% | 75.3% |
| LMArena Expert | 1470 | 1403 |
| Confabulations | — | 14.4% |
| Vectara Hallucination Rate | 10.9% | — |

## Multimodal

- GPT-5.1: 44.0 (#17)
- o3: 40.7 (#35)

| Benchmark | GPT-5.1 | o3 |
|---|---|---|
| LMArena Vision | 1250 | 1214 |
| VPCT | 58.7% | 52% |
| GeoBench | — | 74% |
| LMArena Document | 1403 | — |

## Multilingual

- GPT-5.1: 54.0 (#51)
- o3: 51.9 (#99)

| Benchmark | GPT-5.1 | o3 |
|---|---|---|
| LMArena Non-English | 1431 | 1402 |
| LMArena Chinese | 1495 | 1438 |
| LMArena French | 1451 | 1430 |
| LMArena German | 1439 | 1420 |
| LMArena Japanese | 1453 | 1404 |
| LMArena Korean | 1401 | 1371 |
| LMArena Russian | 1435 | 1407 |
| LMArena Spanish | 1430 | 1395 |

## Instruction Following

- GPT-5.1: 84.0 (#1)
- o3: 73.0 (#122)

| Benchmark | GPT-5.1 | o3 |
|---|---|---|
| IFEval | 93.5% | 86.9% |
| LMArena Instruction Following | 1442 | 1368 |
| LiveBench Instruction Following | 93.3% | — |

## Long Context

- GPT-5.1: 47.6 (#14)
- o3: 53.2 (#6)

| Benchmark | GPT-5.1 | o3 |
|---|---|---|
| CL-bench | 23.7% | 17.8% |
| LMArena Longer Query | 1448 | 1372 |
| Fiction.LiveBench | — | 88.9% |
| CL-bench Life | 17.3% | — |

## Writing & Preference

- GPT-5.1: 64.7 (#53)
- o3: 63.7 (#62)

| Benchmark | GPT-5.1 | o3 |
|---|---|---|
| LMArena Text | 1443 | 1410 |
| LMArena Creative Writing | 1427 | 1359 |
| WildBench | 86.3% | 86.1% |
| LMArena Multi-Turn | 1449 | 1405 |
| Short-Story Creative Writing | — | 83.9% |
| EQ-Bench Creative Writing | — | 1676 |
| LiveBench Language | 80.2% | — |

## FAQ

### Is GPT-5.1 better than o3?

GPT-5.1 is the stronger model overall, scoring 48.8 to 47.4 on the Noometry Index.

### Which is cheaper, GPT-5.1 or o3?

GPT-5.1 is cheaper. It lists at $1.25 per million input tokens and $10 per million output tokens; o3 lists at $2 and $8.

### Is GPT-5.1 or o3 better for coding?

They score almost the same on coding (46.9 vs 47.3); test both on your own repository before choosing.

### Which has the bigger context window?

GPT-5.1 does, with 400K tokens against 200K.

### How many benchmarks do GPT-5.1 and o3 share?

49 benchmarks have published results for both models. GPT-5.1 has 63 scored results on Noometry and o3 has 63.