# Grok-2 (Dec 2024) vs o1

> o1 is the stronger model overall, scoring 40.9 to 33.7 on the Noometry Index.

- Canonical page: https://noometry.com/compare/grok-2-vs-o1
- Last updated: 2026-10-11
- Shared benchmarks: 34

## Summary

- They share 34 benchmarks with published results for both. Grok-2 (Dec 2024) scores higher in 0 categories and o1 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where o1 leads 36.1 to 20.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 11.5% for Grok-2 (Dec 2024) and 73.3% for o1.

## Snapshot

| | Grok-2 (Dec 2024) | o1 |
|---|---|---|
| Provider | xAI | OpenAI |
| Noometry Index | 33.7 | 40.9 |
| Rank | 239 | 143 |
| Context | — | 200K |
| Input $/M | — | $15 |
| Output $/M | — | $60 |
| Weights | Proprietary | Proprietary |

## Coding

- Grok-2 (Dec 2024): 33.3 (#258)
- o1: 46.1 (#70)

| Benchmark | Grok-2 (Dec 2024) | o1 |
|---|---|---|
| WeirdML | 22.2% | 47.6% |
| LiveBench Coding | 46.4% | 69.7% |
| LMArena Coding | 1287 | 1367 |
| Aider Polyglot | — | 61.7% |
| CadEval | — | 56% |
| HumanEval+ | — | 89% |
| MBPP+ | — | 80.2% |

## Agentic & Tool Use

- Grok-2 (Dec 2024): —
- o1: 24.6 (#117)

| Benchmark | Grok-2 (Dec 2024) | o1 |
|---|---|---|
| Cybench | — | 10% |
| METR Time Horizons | — | 51.1% |

## Reasoning

- Grok-2 (Dec 2024): 16.9 (#299)
- o1: 27.9 (#111)

| Benchmark | Grok-2 (Dec 2024) | o1 |
|---|---|---|
| SimpleBench | 22.7% | 41.7% |
| LiveBench Reasoning | 54.8% | 91.6% |
| LMArena Hard Prompts | 1272 | 1371 |
| DTBench | 65.2% | 74.7% |
| LiveBench Data Analysis | 54.5% | 65.5% |
| Epoch Capabilities Index | 130.48 | 141.91 |
| LiveBench | 54.3% | 75.7% |
| ARC-AGI-1 | — | 30.7% |
| Chess Puzzles | — | 15% |
| EnigmaEval | — | 5.7% |
| LMCA | — | 22.3% |

## Math

- Grok-2 (Dec 2024): 20.8 (#284)
- o1: 36.1 (#175)

| Benchmark | Grok-2 (Dec 2024) | o1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 11.5% | 73.3% |
| LiveBench Math | 54.9% | 80.3% |
| LMArena Math | 1283 | 1388 |
| MATH Level 5 | 63.5% | 94.7% |
| FrontierMath (Feb 2025 set) | 0.7% | 9.3% |
| FrontierMath (Tiers 1-3) | — | 14.7% |

## Knowledge

- Grok-2 (Dec 2024): 29.8 (#233)
- o1: 41.5 (#110)

| Benchmark | Grok-2 (Dec 2024) | o1 |
|---|---|---|
| GPQA Diamond | 53.8% | 76.8% |
| Confabulations | 20.1% | 11.7% |
| LMArena Expert | 1254 | 1361 |
| Humanity's Last Exam | — | 8% |
| SimpleQA Verified | — | 41.1% |

## Multimodal

- Grok-2 (Dec 2024): —
- o1: 34.2 (#93)

| Benchmark | Grok-2 (Dec 2024) | o1 |
|---|---|---|
| LMArena Vision | — | 1168 |
| GeoBench | — | 80% |
| VPCT | — | 37% |
| SpatialViz-Bench | — | 41.4% |

## Multilingual

- Grok-2 (Dec 2024): 43.1 (#188)
- o1: 48.6 (#142)

| Benchmark | Grok-2 (Dec 2024) | o1 |
|---|---|---|
| LMArena Non-English | 1282 | 1358 |
| LMArena Chinese | 1289 | 1394 |
| LMArena French | 1318 | 1344 |
| LMArena German | 1287 | 1337 |
| LMArena Japanese | 1244 | 1346 |
| LMArena Korean | 1237 | 1396 |
| LMArena Russian | 1286 | 1356 |
| LMArena Spanish | 1281 | 1345 |

## Instruction Following

- Grok-2 (Dec 2024): 66.9 (#202)
- o1: 74.8 (#86)

| Benchmark | Grok-2 (Dec 2024) | o1 |
|---|---|---|
| LiveBench Instruction Following | 69.6% | 81.5% |
| LMArena Instruction Following | 1270 | 1367 |

## Long Context

- Grok-2 (Dec 2024): 38.8 (#190)
- o1: 50.3 (#9)

| Benchmark | Grok-2 (Dec 2024) | o1 |
|---|---|---|
| LMArena Longer Query | 1276 | 1378 |
| Fiction.LiveBench | — | 83.3% |

## Writing & Preference

- Grok-2 (Dec 2024): 48.6 (#198)
- o1: 55.6 (#144)

| Benchmark | Grok-2 (Dec 2024) | o1 |
|---|---|---|
| LMArena Text | 1305 | 1366 |
| LMArena Creative Writing | 1284 | 1348 |
| Short-Story Creative Writing | 63.6% | 70.2% |
| LMArena Multi-Turn | 1290 | 1369 |
| LiveBench Language | 45.6% | 65.4% |

## FAQ

### Is Grok-2 (Dec 2024) better than o1?

o1 is the stronger model overall, scoring 40.9 to 33.7 on the Noometry Index.

### Is Grok-2 (Dec 2024) or o1 better for coding?

o1 scores higher on coding benchmarks: 46.1 versus 33.3 in the Noometry coding category.

### How many benchmarks do Grok-2 (Dec 2024) and o1 share?

34 benchmarks have published results for both models. Grok-2 (Dec 2024) has 34 scored results on Noometry and o1 has 52.