Model comparison
GPT-4 vs Grok 3
Grok 3 is the stronger model overall, scoring 39.9 to 29.1 on the Noometry Index.
Last verified . 23 shared benchmarks.
Summary
- They share 23 benchmarks with published results for both. GPT-4 scores higher in 1 category and Grok 3 in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Grok 3 leads 46.2 to 18.4.
- The biggest single-benchmark swing is MATH Level 5: 23% for GPT-4 and 88.7% for Grok 3.
Side by side
| GPT-4 | Grok 3 | |
|---|---|---|
| Provider | OpenAI | xAI |
| Noometry Index | 29.1 | 39.9 |
| Released | 2023-03-14 | 2025-04-09 |
| Weights | Proprietary | Proprietary |
| Context window | 8K | — |
| Max output | 8K | — |
| Input $ / M tokens | $30 | — |
| Output $ / M tokens | $60 | — |
| Results tracked | 38 | 40 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 3 leads
GPT-4: 31.6 (#283), Grok 3: 41.9 (#115)
| Benchmark | GPT-4 | Grok 3 |
|---|---|---|
| WeirdML | 12.4% | 37.2% |
| LMArena Coding | 1254 | 1432 |
| Aider Polyglot | — | 53.3% |
| BigCodeBench Instruct | 46% | — |
| BigCodeBench Complete | 57.2% | — |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, Grok 3: 30.5 (#76)
| Benchmark | GPT-4 | Grok 3 |
|---|---|---|
| BALROG | — | 29.5% |
| METR Time Horizons | 36.1% | — |
Reasoning GPT-4 leads
GPT-4: 17.8 (#289), Grok 3: 13.7 (#333)
| Benchmark | GPT-4 | Grok 3 |
|---|---|---|
| LMArena Hard Prompts | 1241 | 1434 |
| Epoch Capabilities Index | 125.89 | 138.33 |
| ARC-AGI-2 | — | 0% |
| SimpleBench | — | 36.1% |
| Kagi LLM Benchmark | — | 61.3% |
| ARC-AGI-1 | — | 5.5% |
| Chess Puzzles | 4% | — |
| Mystery Game Puzzles | 12% | — |
| DTBench | 62.7% | — |
| LMCA | 17.1% | — |
| BIG-Bench Hard | 75.1% | — |
| ForecastBench | 57.8 | — |
| HellaSwag | 95.3% | — |
| WinoGrande | 87.5% | — |
Math Grok 3 leads
GPT-4: 10.8 (#309), Grok 3: 38.0 (#145)
| Benchmark | GPT-4 | Grok 3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.1% | 55.6% |
| LMArena Math | 1269 | 1391 |
| MATH Level 5 | 23% | 88.7% |
| Omni-MATH | — | 46.4% |
| FrontierMath (Feb 2025 set) | — | 3.8% |
| FrontierMath Tier 4 (v1) | — | 0% |
| GSM8K | 92% | — |
Knowledge Grok 3 leads
GPT-4: 18.4 (#282), Grok 3: 46.2 (#82)
| Benchmark | GPT-4 | Grok 3 |
|---|---|---|
| GPQA Diamond | 35.7% | 75.8% |
| LMArena Expert | 1211 | 1421 |
| MMLU-Pro | — | 78.8% |
| Confabulations | — | 14.2% |
| Vectara Hallucination Rate | — | 5.8% |
| GPQA (HELM) | — | 65% |
| MMLU | 86.4% | — |
| TriviaQA | 84.8% | — |
Multilingual Grok 3 leads
GPT-4: 40.6 (#215), Grok 3: 52.3 (#87)
| Benchmark | GPT-4 | Grok 3 |
|---|---|---|
| LMArena Non-English | 1246 | 1410 |
| LMArena Chinese | 1242 | 1448 |
| LMArena French | 1283 | 1460 |
| LMArena German | 1251 | 1431 |
| LMArena Japanese | 1209 | 1387 |
| LMArena Korean | 1184 | 1373 |
| LMArena Russian | 1251 | 1416 |
| LMArena Spanish | 1261 | 1417 |
Instruction Following Grok 3 leads
GPT-4: 65.3 (#222), Grok 3: 75.0 (#73)
| Benchmark | GPT-4 | Grok 3 |
|---|---|---|
| LMArena Instruction Following | 1241 | 1409 |
| IFEval | — | 88.4% |
Long Context Grok 3 leads
GPT-4: 37.7 (#212), Grok 3: 38.7 (#192)
| Benchmark | GPT-4 | Grok 3 |
|---|---|---|
| LMArena Longer Query | 1244 | 1439 |
| Fiction.LiveBench | — | 58.3% |
Writing & Preference Grok 3 leads
GPT-4: 34.9 (#268), Grok 3: 55.8 (#141)
| Benchmark | GPT-4 | Grok 3 |
|---|---|---|
| LMArena Text | 1263 | 1426 |
| LMArena Creative Writing | 1244 | 1414 |
| EQ-Bench Creative Writing | 752 | 1186 |
| LMArena Multi-Turn | 1257 | 1425 |
| Short-Story Creative Writing | — | 76.4% |
| WildBench | — | 84.9% |
Frequently asked questions
Is GPT-4 better than Grok 3?
Grok 3 is the stronger model overall, scoring 39.9 to 29.1 on the Noometry Index.
Is GPT-4 or Grok 3 better for coding?
Grok 3 scores higher on coding benchmarks: 41.9 versus 31.6 in the Noometry coding category.
How many benchmarks do GPT-4 and Grok 3 share?
23 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Grok 3 has 40.