Model comparison
Grok 3 vs Grok 4
Grok 4 is the stronger model overall, scoring 48.1 to 39.9 on the Noometry Index.
Last verified . 37 shared benchmarks.
Summary
- They share 37 benchmarks with published results for both. Grok 3 scores higher in 1 category and Grok 4 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in long context, where Grok 4 leads 63.1 to 38.7.
- The biggest single-benchmark swing is ARC-AGI-1: 5.5% for Grok 3 and 66.7% for Grok 4.
Side by side
| Grok 3 | Grok 4 | |
|---|---|---|
| Provider | xAI | xAI |
| Noometry Index | 39.9 | 48.1 |
| Released | 2025-04-09 | 2025-07-09 |
| Weights | Proprietary | Proprietary |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 40 | 48 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 4 leads
Grok 3: 41.9 (#115), Grok 4: 50.3 (#46)
| Benchmark | Grok 3 | Grok 4 |
|---|---|---|
| Aider Polyglot | 53.3% | 79.6% |
| WeirdML | 37.2% | 45.7% |
| LMArena Coding | 1432 | 1408 |
Agentic & Tool Use Grok 4 leads
Grok 3: 30.5 (#76), Grok 4: 32.3 (#68)
| Benchmark | Grok 3 | Grok 4 |
|---|---|---|
| BALROG | 29.5% | 43.6% |
| Terminal-Bench | — | 27.2% |
| Berkeley Function Calling Leaderboard | — | 63% |
| GDPval | — | 21.1% |
| Cybench | — | 43% |
| DeepResearch Bench | — | 47.3% |
| LMArena Search | — | 1142 |
| METR Time Horizons | — | 66.6% |
Reasoning Grok 4 leads
Grok 3: 13.7 (#333), Grok 4: 36.7 (#65)
| Benchmark | Grok 3 | Grok 4 |
|---|---|---|
| ARC-AGI-2 | 0% | 16% |
| SimpleBench | 36.1% | 60.5% |
| Kagi LLM Benchmark | 61.3% | 73.6% |
| ARC-AGI-1 | 5.5% | 66.7% |
| LMArena Hard Prompts | 1434 | 1409 |
| Epoch Capabilities Index | 138.33 | 146.44 |
| Chess Puzzles | — | 28% |
| ForecastBench | — | 60.9 |
Math Grok 4 leads
Grok 3: 38.0 (#145), Grok 4: 48.4 (#64)
| Benchmark | Grok 3 | Grok 4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 55.6% | 84% |
| Omni-MATH | 46.4% | 60.3% |
| LMArena Math | 1391 | 1422 |
| FrontierMath (Feb 2025 set) | 3.8% | 19.7% |
| FrontierMath Tier 4 (v1) | 0% | 2.1% |
| MATH Level 5 | 88.7% | — |
Knowledge Grok 4 leads
Grok 3: 46.2 (#82), Grok 4: 53.8 (#55)
| Benchmark | Grok 3 | Grok 4 |
|---|---|---|
| GPQA Diamond | 75.8% | 87% |
| MMLU-Pro | 78.8% | 85.1% |
| Confabulations | 14.2% | 12.4% |
| GPQA (HELM) | 65% | 72.7% |
| LMArena Expert | 1421 | 1415 |
| Vectara Hallucination Rate | 5.8% | — |
Multimodal Not comparable
Grok 3: —, Grok 4: 33.7 (#94)
| Benchmark | Grok 3 | Grok 4 |
|---|---|---|
| LMArena Vision | — | 1210 |
| GeoBench | — | 45% |
Multilingual Too close to call
Grok 3: 52.3 (#87), Grok 4: 51.8 (#103)
| Benchmark | Grok 3 | Grok 4 |
|---|---|---|
| LMArena Non-English | 1410 | 1403 |
| LMArena Chinese | 1448 | 1427 |
| LMArena French | 1460 | 1418 |
| LMArena German | 1431 | 1429 |
| LMArena Japanese | 1387 | 1394 |
| LMArena Korean | 1373 | 1377 |
| LMArena Russian | 1416 | 1410 |
| LMArena Spanish | 1417 | 1420 |
Instruction Following Grok 4 leads
Grok 3: 75.0 (#73), Grok 4: 79.2 (#5)
| Benchmark | Grok 3 | Grok 4 |
|---|---|---|
| IFEval | 88.4% | 94.9% |
| LMArena Instruction Following | 1409 | 1387 |
Long Context Grok 4 leads
Grok 3: 38.7 (#192), Grok 4: 63.1 (#4)
| Benchmark | Grok 3 | Grok 4 |
|---|---|---|
| Fiction.LiveBench | 58.3% | 94.4% |
| LMArena Longer Query | 1439 | 1409 |
Writing & Preference Grok 4 leads
Grok 3: 55.8 (#141), Grok 4: 58.5 (#116)
| Benchmark | Grok 3 | Grok 4 |
|---|---|---|
| LMArena Text | 1426 | 1411 |
| LMArena Creative Writing | 1414 | 1397 |
| Short-Story Creative Writing | 76.4% | 76.9% |
| WildBench | 84.9% | 79.7% |
| LMArena Multi-Turn | 1425 | 1416 |
| EQ-Bench Creative Writing | 1186 | — |
Frequently asked questions
Is Grok 3 better than Grok 4?
Grok 4 is the stronger model overall, scoring 48.1 to 39.9 on the Noometry Index.
Is Grok 3 or Grok 4 better for coding?
Grok 4 scores higher on coding benchmarks: 50.3 versus 41.9 in the Noometry coding category.
How many benchmarks do Grok 3 and Grok 4 share?
37 benchmarks have published results for both models. Grok 3 has 40 scored results on Noometry and Grok 4 has 48.