# GPT-5.2 vs Grok 4.5

> GPT-5.2 and Grok 4.5 score almost the same on the Noometry Index (54.1 vs 55.0), so choose on price, context window or the category you care about most.

- Canonical page: https://noometry.com/compare/gpt-5-2-vs-grok-4-5
- Last updated: 2026-10-10
- Shared benchmarks: 42

## Summary

- They share 42 benchmarks with published results for both. GPT-5.2 scores higher in 2 categories and Grok 4.5 in 8 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in multimodal, where GPT-5.2 leads 51.3 to 37.6.
- The biggest single-benchmark swing is WeirdML: 72.2% for GPT-5.2 and 46.4% for Grok 4.5.
- Grok 4.5 is cheaper at $2 / $6 per million input/output tokens, against $1.75 / $14 for GPT-5.2.
- Grok 4.5 accepts more context: 500K tokens versus 400K.

## Snapshot

| | GPT-5.2 | Grok 4.5 |
|---|---|---|
| Provider | OpenAI | xAI |
| Noometry Index | 54.1 | 55.0 |
| Rank | 34 | 25 |
| Context | 400K | 500K |
| Input $/M | $1.75 | $2 |
| Output $/M | $14 | $6 |
| Weights | Proprietary | Proprietary |

## Coding

- GPT-5.2: 51.6 (#37)
- Grok 4.5: 52.2 (#35)

| Benchmark | GPT-5.2 | Grok 4.5 |
|---|---|---|
| LMArena WebDev | 1416 | 1553 |
| WeirdML | 72.2% | 46.4% |
| LMArena Coding | 1447 | 1474 |
| ALE-Bench | 1,294 | 1,309 |
| SWE-bench Verified | 73.8% | — |
| DeepSWE | — | 53.8% |
| FrontierCode | — | 42.4% |
| SWE-bench Verified (bash only) | 72.8% | — |
| SWE-bench Multilingual | 66.7% | — |
| SciCode | — | 54.1% |
| GSO | 27.4% | — |
| AlgoTune | 2.05 | — |

## Agentic & Tool Use

- GPT-5.2: 40.2 (#24)
- Grok 4.5: 44.4 (#17)

| Benchmark | GPT-5.2 | Grok 4.5 |
|---|---|---|
| τ²-bench Banking | 32.2% | 47.9% |
| LMArena Search | 1207 | 1213 |
| Vending-Bench 2 | 3,591 | 3,887 |
| Terminal-Bench | 64.9% | — |
| APEX-Agents | — | 56.2% |
| Berkeley Function Calling Leaderboard | 55.9% | — |
| GDPval | 49.7% | — |
| Remote Labor Index | 2.5% | — |
| τ²-bench Airline | 83% | — |
| τ²-bench Retail | 81.6% | — |
| τ²-bench Telecom | 89.7% | — |
| DeepResearch Bench | 41.1% | — |
| PostTrainBench | — | 23.4% |
| GBAEval | — | 65.4% |
| GDP.pdf | — | 14% |
| METR Time Horizons | 75.3% | — |

## Reasoning

- GPT-5.2: 50.2 (#35)
- Grok 4.5: 56.1 (#25)

| Benchmark | GPT-5.2 | Grok 4.5 |
|---|---|---|
| ARC-AGI-2 | 52.9% | 52.6% |
| SimpleBench | 45.8% | 70% |
| Kagi LLM Benchmark | 73.3% | 83.5% |
| NYT Connections (extended) | 83.6% | 79.9% |
| ARC-AGI-1 | 86.2% | 87.2% |
| Chess Puzzles | 49% | 36% |
| LMArena Hard Prompts | 1445 | 1462 |
| DTBench | 90.9% | 96.5% |
| LMCA | 43.9% | 45.2% |
| Epoch Capabilities Index | 153.45 | 153.92 |
| CritPt | — | 15.4% |
| EnigmaEval | 10.4% | — |
| EBR-Bench | 23% | — |
| Mystery Game Puzzles | 23% | — |
| Surface Evolver Bench | — | 74.4% |
| ForecastBench | 60.1 | — |

## Math

- GPT-5.2: 60.0 (#38)
- Grok 4.5: 60.9 (#35)

| Benchmark | GPT-5.2 | Grok 4.5 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 67.4% | 57.2% |
| FrontierMath Tier 4 | 31.7% | 24.4% |
| OTIS Mock AIME 2024-2025 | 96.1% | 97.8% |
| ProofBench | 15% | 31% |
| LMArena Math | 1440 | 1459 |
| MathArena Final-Answer Competitions | 72% | — |
| FrontierMath (Feb 2025 set) | 40.7% | — |
| FrontierMath Tier 4 (v1) | 18.8% | — |

## Knowledge

- GPT-5.2: 59.3 (#32)
- Grok 4.5: 62.3 (#24)

| Benchmark | GPT-5.2 | Grok 4.5 |
|---|---|---|
| GPQA Diamond | 91.4% | 93.4% |
| SimpleQA Verified | 37.1% | 48.3% |
| LMArena Expert | 1445 | 1466 |
| Humanity's Last Exam | 27.8% | — |
| Vectara Hallucination Rate | 8.4% | — |

## Multimodal

- GPT-5.2: 51.3 (#7)
- Grok 4.5: 37.6 (#72)

| Benchmark | GPT-5.2 | Grok 4.5 |
|---|---|---|
| LMArena Vision | 1268 | 1288 |
| Furniture Assembly | 38.3% | 22.5% |
| LMArena Document | 1405 | 1452 |
| VPCT | 84% | — |
| Blueprint-Bench 2 | — | 27.3% |

## Multilingual

- GPT-5.2: 53.4 (#67)
- Grok 4.5: 54.4 (#42)

| Benchmark | GPT-5.2 | Grok 4.5 |
|---|---|---|
| LMArena Non-English | 1425 | 1440 |
| LMArena Chinese | 1460 | 1496 |
| LMArena French | 1455 | 1456 |
| LMArena German | 1448 | 1446 |
| LMArena Japanese | 1420 | 1428 |
| LMArena Korean | 1392 | 1404 |
| LMArena Russian | 1440 | 1448 |
| LMArena Spanish | 1433 | 1450 |

## Instruction Following

- GPT-5.2: 74.7 (#89)
- Grok 4.5: 76.0 (#48)

| Benchmark | GPT-5.2 | Grok 4.5 |
|---|---|---|
| LMArena Instruction Following | 1417 | 1446 |

## Long Context

- GPT-5.2: 44.0 (#78)
- Grok 4.5: 44.8 (#56)

| Benchmark | GPT-5.2 | Grok 4.5 |
|---|---|---|
| LMArena Longer Query | 1428 | 1463 |
| CL-bench | 18.2% | — |

## Writing & Preference

- GPT-5.2: 66.8 (#32)
- Grok 4.5: 65.8 (#42)

| Benchmark | GPT-5.2 | Grok 4.5 |
|---|---|---|
| LMArena Text | 1439 | 1448 |
| LMArena Creative Writing | 1401 | 1442 |
| EQ-Bench Creative Writing | 1703 | 1579 |
| LMArena Multi-Turn | 1458 | 1456 |

## FAQ

### Is GPT-5.2 better than Grok 4.5?

GPT-5.2 and Grok 4.5 score almost the same on the Noometry Index (54.1 vs 55.0), so choose on price, context window or the category you care about most.

### Which is cheaper, GPT-5.2 or Grok 4.5?

Grok 4.5 is cheaper. It lists at $2 per million input tokens and $6 per million output tokens; GPT-5.2 lists at $1.75 and $14.

### Is GPT-5.2 or Grok 4.5 better for coding?

They score almost the same on coding (51.6 vs 52.2); test both on your own repository before choosing.

### Which has the bigger context window?

Grok 4.5 does, with 500K tokens against 400K.

### How many benchmarks do GPT-5.2 and Grok 4.5 share?

42 benchmarks have published results for both models. GPT-5.2 has 67 scored results on Noometry and Grok 4.5 has 52.