# GPT-5.5 vs o3

> GPT-5.5 is the stronger model overall, scoring 63.4 to 47.5 on the Noometry Index. o3 costs 3.2× less per token, which makes it the better buy when GPT-5.5's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/gpt-5-5-vs-o3
- Last updated: 2026-10-11
- Shared benchmarks: 42

## Summary

- They share 42 benchmarks with published results for both. GPT-5.5 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-5.5 leads 72.8 to 32.0.
- The biggest single-benchmark swing is ARC-AGI-2: 85% for GPT-5.5 and 6.5% for o3.
- o3 is cheaper at $2 / $8 per million input/output tokens, against $5 / $30 for GPT-5.5.
- GPT-5.5 accepts more context: 1.05M tokens versus 200K.

## Snapshot

| | GPT-5.5 | o3 |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 63.4 | 47.5 |
| Rank | 9 | 61 |
| Context | 1.05M | 200K |
| Input $/M | $5 | $2 |
| Output $/M | $30 | $8 |
| Weights | Proprietary | Proprietary |

## Coding

- GPT-5.5: 58.2 (#17)
- o3: 46.8 (#64)

| Benchmark | GPT-5.5 | o3 |
|---|---|---|
| SWE-bench Verified | 80.6% | 62.3% |
| GSO | 40.2% | 8.8% |
| WeirdML | 84.9% | 52.4% |
| LMArena Coding | 1494 | 1408 |
| ALE-Bench | 1,943 | 933.55 |
| DeepSWE | 67% | — |
| FrontierCode | 43% | — |
| SWE-bench Verified (bash only) | — | 58.4% |
| Aider Polyglot | — | 81.3% |
| LMArena WebDev | 1513 | — |
| SciCode | 56.1% | — |
| MirrorCode | 10% | — |
| CadEval | — | 74% |

## Agentic & Tool Use

- GPT-5.5: 50.7 (#6)
- o3: 34.5 (#44)

| Benchmark | GPT-5.5 | o3 |
|---|---|---|
| DeepResearch Bench | 54% | 45.2% |
| LMArena Search | 1242 | 1144 |
| Terminal-Bench | 84.7% | — |
| APEX-Agents | 55.1% | — |
| Berkeley Function Calling Leaderboard | — | 63% |
| OSWorld 2.0 | 13% | — |
| GDPval | — | 30.8% |
| Remote Labor Index | 6.3% | — |
| τ²-bench Banking | 44.6% | — |
| OSWorld | — | 23% |
| PostTrainBench | 27.2% | — |
| ExploitBench | 47.4% | — |
| GBAEval | 53.2% | — |
| GDP.pdf | 26% | — |
| METR Time Horizons | — | 65.4% |
| Vending-Bench 2 | 7,524 | — |

## Reasoning

- GPT-5.5: 72.8 (#11)
- o3: 32.0 (#78)

| Benchmark | GPT-5.5 | o3 |
|---|---|---|
| ARC-AGI-2 | 85% | 6.5% |
| SimpleBench | 69% | 53.1% |
| Kagi LLM Benchmark | 88.8% | 67.6% |
| ARC-AGI-1 | 95% | 60.8% |
| CritPt | 27.1% | 1.4% |
| Chess Puzzles | 54% | 38% |
| LMArena Hard Prompts | 1489 | 1402 |
| Mystery Game Puzzles | 56% | 29% |
| DTBench | 96% | 84.8% |
| LMCA | 54.3% | 39.7% |
| Epoch Capabilities Index | 159.1 | 146.86 |
| ForecastBench | 60.6 | 62.5 |
| NYT Connections (extended) | 96.2% | — |
| EnigmaEval | — | 13.1% |
| EBR-Bench | 34.3% | — |
| Surface Evolver Bench | 88.1% | — |
| Bench to the Future 3 | 0.14 | — |

## Math

- GPT-5.5: 81.7 (#11)
- o3: 50.2 (#58)

| Benchmark | GPT-5.5 | o3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 85.3% | 33.3% |
| OTIS Mock AIME 2024-2025 | 100% | 84.4% |
| LMArena Math | 1486 | 1426 |
| FrontierMath (Feb 2025 set) | 51.7% | 18.7% |
| FrontierMath Tier 4 (v1) | 35.4% | 2.1% |
| FrontierMath Tier 4 | 72.5% | — |
| MathArena Final-Answer Competitions | 94.3% | — |
| ProofBench | 50% | — |
| Omni-MATH | — | 71.4% |
| MATH Level 5 | — | 97.8% |
| FrontierMath Erdős | 0% | — |

## Knowledge

- GPT-5.5: 64.4 (#17)
- o3: 54.6 (#52)

| Benchmark | GPT-5.5 | o3 |
|---|---|---|
| GPQA Diamond | 94% | 81.8% |
| SimpleQA Verified | 63% | 49.4% |
| LMArena Expert | 1508 | 1402 |
| Humanity's Last Exam | — | 20.3% |
| MMLU-Pro | — | 85.9% |
| Confabulations | — | 14.4% |
| Vectara Hallucination Rate | 9.3% | — |
| GPQA (HELM) | — | 75.3% |

## Multimodal

- GPT-5.5: 46.9 (#12)
- o3: 41.4 (#36)

| Benchmark | GPT-5.5 | o3 |
|---|---|---|
| LMArena Vision | 1297 | 1214 |
| GeoBench | — | 74% |
| VPCT | — | 52% |
| Blueprint-Bench 2 | 36.2% | — |
| Furniture Assembly | 44.2% | — |
| LMArena Document | 1486 | — |

## Multilingual

- GPT-5.5: 56.4 (#20)
- o3: 51.7 (#105)

| Benchmark | GPT-5.5 | o3 |
|---|---|---|
| LMArena Non-English | 1467 | 1401 |
| LMArena Chinese | 1533 | 1437 |
| LMArena French | 1486 | 1430 |
| LMArena German | 1480 | 1420 |
| LMArena Japanese | 1498 | 1403 |
| LMArena Korean | 1460 | 1370 |
| LMArena Russian | 1473 | 1406 |
| LMArena Spanish | 1468 | 1395 |

## Instruction Following

- GPT-5.5: 77.5 (#18)
- o3: 72.8 (#127)

| Benchmark | GPT-5.5 | o3 |
|---|---|---|
| LMArena Instruction Following | 1479 | 1368 |
| IFEval | — | 86.9% |

## Long Context

- GPT-5.5: 48.3 (#12)
- o3: 53.3 (#6)

| Benchmark | GPT-5.5 | o3 |
|---|---|---|
| LMArena Longer Query | 1484 | 1372 |
| Fiction.LiveBench | — | 88.9% |
| CL-bench | — | 17.8% |
| CL-bench Life | 22.2% | — |

## Writing & Preference

- GPT-5.5: 72.7 (#13)
- o3: 63.5 (#64)

| Benchmark | GPT-5.5 | o3 |
|---|---|---|
| LMArena Text | 1472 | 1410 |
| LMArena Creative Writing | 1455 | 1359 |
| EQ-Bench Creative Writing | 1844 | 1676 |
| LMArena Multi-Turn | 1476 | 1405 |
| Short-Story Creative Writing | — | 83.9% |
| WildBench | — | 86.1% |
| EQ-Bench 4 | 1315 | — |

## FAQ

### Is GPT-5.5 better than o3?

GPT-5.5 is the stronger model overall, scoring 63.4 to 47.5 on the Noometry Index. o3 costs 3.2× less per token, which makes it the better buy when GPT-5.5's lead doesn't matter for your workload.

### Which is cheaper, GPT-5.5 or o3?

o3 is cheaper. It lists at $2 per million input tokens and $8 per million output tokens; GPT-5.5 lists at $5 and $30.

### Is GPT-5.5 or o3 better for coding?

GPT-5.5 scores higher on coding benchmarks: 58.2 versus 46.8 in the Noometry coding category.

### Which has the bigger context window?

GPT-5.5 does, with 1.05M tokens against 200K.

### How many benchmarks do GPT-5.5 and o3 share?

42 benchmarks have published results for both models. GPT-5.5 has 71 scored results on Noometry and o3 has 63.