Model comparison
Grok-2 (Dec 2024) vs Grok 3
Grok 3 is the stronger model overall, scoring 39.9 to 33.7 on the Noometry Index.
Last verified . 26 shared benchmarks.
Summary
- They share 26 benchmarks with published results for both. Grok-2 (Dec 2024) scores higher in 2 categories and Grok 3 in 6 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Grok 3 leads 38.0 to 20.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 11.5% for Grok-2 (Dec 2024) and 55.6% for Grok 3.
Side by side
| Grok-2 (Dec 2024) | Grok 3 | |
|---|---|---|
| Provider | xAI | xAI |
| Noometry Index | 33.7 | 39.9 |
| Released | 2024-08-13 | 2025-04-09 |
| Weights | Proprietary | Proprietary |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 34 | 40 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 3 leads
Grok-2 (Dec 2024): 33.3 (#258), Grok 3: 41.9 (#115)
| Benchmark | Grok-2 (Dec 2024) | Grok 3 |
|---|---|---|
| WeirdML | 22.2% | 37.2% |
| LMArena Coding | 1287 | 1432 |
| Aider Polyglot | — | 53.3% |
| LiveBench Coding | 46.4% | — |
Agentic & Tool Use Not comparable
Grok-2 (Dec 2024): —, Grok 3: 30.5 (#76)
| Benchmark | Grok-2 (Dec 2024) | Grok 3 |
|---|---|---|
| BALROG | — | 29.5% |
Reasoning Grok-2 (Dec 2024) leads
Grok-2 (Dec 2024): 16.9 (#299), Grok 3: 13.7 (#333)
| Benchmark | Grok-2 (Dec 2024) | Grok 3 |
|---|---|---|
| SimpleBench | 22.7% | 36.1% |
| LMArena Hard Prompts | 1272 | 1434 |
| Epoch Capabilities Index | 130.48 | 138.33 |
| ARC-AGI-2 | — | 0% |
| Kagi LLM Benchmark | — | 61.3% |
| ARC-AGI-1 | — | 5.5% |
| LiveBench Reasoning | 54.8% | — |
| DTBench | 65.2% | — |
| LiveBench Data Analysis | 54.5% | — |
| LiveBench | 54.3% | — |
Math Grok 3 leads
Grok-2 (Dec 2024): 20.8 (#284), Grok 3: 38.0 (#145)
| Benchmark | Grok-2 (Dec 2024) | Grok 3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 11.5% | 55.6% |
| LMArena Math | 1283 | 1391 |
| MATH Level 5 | 63.5% | 88.7% |
| FrontierMath (Feb 2025 set) | 0.7% | 3.8% |
| Omni-MATH | — | 46.4% |
| LiveBench Math | 54.9% | — |
| FrontierMath Tier 4 (v1) | — | 0% |
Knowledge Grok 3 leads
Grok-2 (Dec 2024): 29.8 (#233), Grok 3: 46.2 (#82)
| Benchmark | Grok-2 (Dec 2024) | Grok 3 |
|---|---|---|
| GPQA Diamond | 53.8% | 75.8% |
| Confabulations | 20.1% | 14.2% |
| LMArena Expert | 1254 | 1421 |
| MMLU-Pro | — | 78.8% |
| Vectara Hallucination Rate | — | 5.8% |
| GPQA (HELM) | — | 65% |
Multilingual Grok 3 leads
Grok-2 (Dec 2024): 43.1 (#188), Grok 3: 52.3 (#87)
| Benchmark | Grok-2 (Dec 2024) | Grok 3 |
|---|---|---|
| LMArena Non-English | 1282 | 1410 |
| LMArena Chinese | 1289 | 1448 |
| LMArena French | 1318 | 1460 |
| LMArena German | 1287 | 1431 |
| LMArena Japanese | 1244 | 1387 |
| LMArena Korean | 1237 | 1373 |
| LMArena Russian | 1286 | 1416 |
| LMArena Spanish | 1281 | 1417 |
Instruction Following Grok 3 leads
Grok-2 (Dec 2024): 66.9 (#202), Grok 3: 75.0 (#73)
| Benchmark | Grok-2 (Dec 2024) | Grok 3 |
|---|---|---|
| LMArena Instruction Following | 1270 | 1409 |
| LiveBench Instruction Following | 69.6% | — |
| IFEval | — | 88.4% |
Long Context Too close to call
Grok-2 (Dec 2024): 38.8 (#190), Grok 3: 38.7 (#192)
| Benchmark | Grok-2 (Dec 2024) | Grok 3 |
|---|---|---|
| LMArena Longer Query | 1276 | 1439 |
| Fiction.LiveBench | — | 58.3% |
Writing & Preference Grok 3 leads
Grok-2 (Dec 2024): 48.6 (#198), Grok 3: 55.8 (#141)
| Benchmark | Grok-2 (Dec 2024) | Grok 3 |
|---|---|---|
| LMArena Text | 1305 | 1426 |
| LMArena Creative Writing | 1284 | 1414 |
| Short-Story Creative Writing | 63.6% | 76.4% |
| LMArena Multi-Turn | 1290 | 1425 |
| EQ-Bench Creative Writing | — | 1186 |
| WildBench | — | 84.9% |
| LiveBench Language | 45.6% | — |
Frequently asked questions
Is Grok-2 (Dec 2024) better than Grok 3?
Grok 3 is the stronger model overall, scoring 39.9 to 33.7 on the Noometry Index.
Is Grok-2 (Dec 2024) or Grok 3 better for coding?
Grok 3 scores higher on coding benchmarks: 41.9 versus 33.3 in the Noometry coding category.
How many benchmarks do Grok-2 (Dec 2024) and Grok 3 share?
26 benchmarks have published results for both models. Grok-2 (Dec 2024) has 34 scored results on Noometry and Grok 3 has 40.