# GPT-5 vs GPT-5.2

> GPT-5.2 is the stronger model overall, scoring 54.1 to 50.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/gpt-5-vs-gpt-5-2
- Last updated: 2026-10-10
- Shared benchmarks: 55

## Summary

- They share 55 benchmarks with published results for both. GPT-5 scores higher in 1 category and GPT-5.2 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in long context, where GPT-5 leads 69.5 to 44.0.
- The biggest single-benchmark swing is ARC-AGI-2: 9.9% for GPT-5 and 52.9% for GPT-5.2.
- GPT-5 is cheaper at $1.25 / $10 per million input/output tokens, against $1.75 / $14 for GPT-5.2.

## Snapshot

| | GPT-5 | GPT-5.2 |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 50.9 | 54.1 |
| Rank | 37 | 31 |
| Context | 400K | 400K |
| Input $/M | $1.25 | $1.75 |
| Output $/M | $10 | $14 |
| Weights | Proprietary | Proprietary |

## Coding

- GPT-5: 50.7 (#39)
- GPT-5.2: 52.2 (#32)

| Benchmark | GPT-5 | GPT-5.2 |
|---|---|---|
| SWE-bench Verified | 73.6% | 73.8% |
| SWE-bench Verified (bash only) | 65% | 72.8% |
| LMArena WebDev | 1417 | 1415 |
| GSO | 6.9% | 27.4% |
| WeirdML | 60.7% | 72.2% |
| LMArena Coding | 1436 | 1447 |
| ALE-Bench | 1,162 | 1,294 |
| AlgoTune | 1.67 | 2.05 |
| Aider Polyglot | 88% | — |
| SWE-bench Multilingual | — | 66.7% |
| SciCode | 42.9% | — |

## Agentic & Tool Use

- GPT-5: 33.0 (#53)
- GPT-5.2: 40.0 (#24)

| Benchmark | GPT-5 | GPT-5.2 |
|---|---|---|
| Terminal-Bench | 49.6% | 64.9% |
| GDPval | 34.8% | 49.7% |
| Remote Labor Index | 1.7% | 2.5% |
| DeepResearch Bench | 49.6% | 41.1% |
| LMArena Search | 1133 | 1207 |
| METR Time Horizons | 69.6% | 75.3% |
| Berkeley Function Calling Leaderboard | — | 55.9% |
| τ²-bench Airline | — | 83% |
| τ²-bench Banking | — | 32.2% |
| τ²-bench Retail | — | 81.6% |
| τ²-bench Telecom | — | 89.7% |
| BALROG | 32.8% | — |
| Vending-Bench 2 | — | 3,591 |

## Reasoning

- GPT-5: 38.3 (#55)
- GPT-5.2: 50.1 (#33)

| Benchmark | GPT-5 | GPT-5.2 |
|---|---|---|
| ARC-AGI-2 | 9.9% | 52.9% |
| SimpleBench | 56.7% | 45.8% |
| Kagi LLM Benchmark | 72.7% | 73.3% |
| ARC-AGI-1 | 65.7% | 86.2% |
| Chess Puzzles | 37% | 49% |
| EnigmaEval | 10.5% | 10.4% |
| EBR-Bench | 12.7% | 23% |
| LMArena Hard Prompts | 1415 | 1445 |
| Mystery Game Puzzles | 23% | 23% |
| DTBench | 90.7% | 90.9% |
| LMCA | 40% | 43.9% |
| Epoch Capabilities Index | 150 | 153.45 |
| ForecastBench | 61.4 | 60.1 |
| NYT Connections (extended) | — | 83.6% |
| CritPt | 12.6% | — |

## Math

- GPT-5: 54.6 (#36)
- GPT-5.2: 59.8 (#32)

| Benchmark | GPT-5 | GPT-5.2 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 55.4% | 67.4% |
| FrontierMath Tier 4 | 22% | 31.7% |
| OTIS Mock AIME 2024-2025 | 91.4% | 96.1% |
| ProofBench | 18% | 15% |
| LMArena Math | 1406 | 1439 |
| FrontierMath (Feb 2025 set) | 32.4% | 40.7% |
| FrontierMath Tier 4 (v1) | 12.5% | 18.8% |
| MathArena Final-Answer Competitions | — | 72% |
| Omni-MATH | 64.7% | — |
| MATH Level 5 | 98.1% | — |

## Knowledge

- GPT-5: 56.5 (#37)
- GPT-5.2: 59.2 (#29)

| Benchmark | GPT-5 | GPT-5.2 |
|---|---|---|
| GPQA Diamond | 86.2% | 91.4% |
| Humanity's Last Exam | 25.3% | 27.8% |
| SimpleQA Verified | 50.1% | 37.1% |
| Vectara Hallucination Rate | 14.7% | 8.4% |
| LMArena Expert | 1420 | 1445 |
| MMLU-Pro | 86.3% | — |
| Confabulations | 10.3% | — |
| GPQA (HELM) | 79.2% | — |

## Multimodal

- GPT-5: 46.1 (#12)
- GPT-5.2: 50.9 (#7)

| Benchmark | GPT-5 | GPT-5.2 |
|---|---|---|
| LMArena Vision | 1232 | 1268 |
| VPCT | 66% | 84% |
| GeoBench | 81% | — |
| Furniture Assembly | — | 38.3% |
| LMArena Document | — | 1405 |

## Multilingual

- GPT-5: 51.5 (#105)
- GPT-5.2: 53.6 (#63)

| Benchmark | GPT-5 | GPT-5.2 |
|---|---|---|
| LMArena Non-English | 1397 | 1425 |
| LMArena Chinese | 1423 | 1460 |
| LMArena French | 1409 | 1456 |
| LMArena German | 1415 | 1448 |
| LMArena Japanese | 1411 | 1421 |
| LMArena Korean | 1361 | 1390 |
| LMArena Russian | 1406 | 1440 |
| LMArena Spanish | 1398 | 1433 |

## Instruction Following

- GPT-5: 74.0 (#107)
- GPT-5.2: 74.9 (#83)

| Benchmark | GPT-5 | GPT-5.2 |
|---|---|---|
| LMArena Instruction Following | 1388 | 1417 |
| IFEval | 87.5% | — |

## Long Context

- GPT-5: 69.5 (#2)
- GPT-5.2: 44.0 (#74)

| Benchmark | GPT-5 | GPT-5.2 |
|---|---|---|
| LMArena Longer Query | 1399 | 1428 |
| Fiction.LiveBench | 97.2% | — |
| CL-bench | — | 18.2% |

## Writing & Preference

- GPT-5: 63.5 (#63)
- GPT-5.2: 67.0 (#30)

| Benchmark | GPT-5 | GPT-5.2 |
|---|---|---|
| LMArena Text | 1406 | 1439 |
| LMArena Creative Writing | 1365 | 1402 |
| EQ-Bench Creative Writing | 1627 | 1703 |
| LMArena Multi-Turn | 1426 | 1458 |
| Short-Story Creative Writing | 86% | — |
| WildBench | 85.7% | — |

## FAQ

### Is GPT-5 better than GPT-5.2?

GPT-5.2 is the stronger model overall, scoring 54.1 to 50.9 on the Noometry Index.

### Which is cheaper, GPT-5 or GPT-5.2?

GPT-5 is cheaper. It lists at $1.25 per million input tokens and $10 per million output tokens; GPT-5.2 lists at $1.75 and $14.

### Is GPT-5 or GPT-5.2 better for coding?

GPT-5.2 scores higher on coding benchmarks: 52.2 versus 50.7 in the Noometry coding category.

### Which has the bigger context window?

Both accept 400K tokens.

### How many benchmarks do GPT-5 and GPT-5.2 share?

55 benchmarks have published results for both models. GPT-5 has 69 scored results on Noometry and GPT-5.2 has 67.