Model comparison
GPT-4 vs Grok 4 Fast
Grok 4 Fast is the stronger model overall, scoring 39.4 to 29.1 on the Noometry Index.
Last verified . 21 shared benchmarks.
Summary
- They share 21 benchmarks with published results for both. GPT-4 scores higher in 0 categories and Grok 4 Fast in 8 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Grok 4 Fast leads 38.9 to 10.8.
- The biggest single-benchmark swing is WeirdML: 12.4% for GPT-4 and 42.9% for Grok 4 Fast.
Side by side
| GPT-4 | Grok 4 Fast | |
|---|---|---|
| Provider | OpenAI | xAI |
| Noometry Index | 29.1 | 39.4 |
| Released | 2023-03-14 | 2025-09-19 |
| Weights | Proprietary | Proprietary |
| Context window | 8K | — |
| Max output | 8K | — |
| Input $ / M tokens | $30 | — |
| Output $ / M tokens | $60 | — |
| Results tracked | 38 | 30 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
GPT-4: 31.6 (#283), Grok 4 Fast: 32.5 (#271)
| Benchmark | GPT-4 | Grok 4 Fast |
|---|---|---|
| WeirdML | 12.4% | 42.9% |
| LMArena Coding | 1254 | 1429 |
| LMArena WebDev | — | 1159 |
| BigCodeBench Instruct | 46% | — |
| BigCodeBench Complete | 57.2% | — |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, Grok 4 Fast: 29.5 (#86)
| Benchmark | GPT-4 | Grok 4 Fast |
|---|---|---|
| τ²-bench Banking | — | 15.7% |
| Cybench | — | 30% |
| LMArena Search | — | 1171 |
| METR Time Horizons | 36.1% | — |
Reasoning Grok 4 Fast leads
GPT-4: 17.8 (#289), Grok 4 Fast: 22.2 (#201)
| Benchmark | GPT-4 | Grok 4 Fast |
|---|---|---|
| LMArena Hard Prompts | 1241 | 1412 |
| DTBench | 62.7% | 82.7% |
| Epoch Capabilities Index | 125.89 | 144.2 |
| ForecastBench | 57.8 | 60.5 |
| ARC-AGI-2 | — | 5.3% |
| Kagi LLM Benchmark | — | 66.1% |
| ARC-AGI-1 | — | 48.5% |
| Chess Puzzles | 4% | — |
| Mystery Game Puzzles | 12% | — |
| LMCA | 17.1% | — |
| BIG-Bench Hard | 75.1% | — |
| HellaSwag | 95.3% | — |
| WinoGrande | 87.5% | — |
Math Grok 4 Fast leads
GPT-4: 10.8 (#309), Grok 4 Fast: 38.9 (#123)
| Benchmark | GPT-4 | Grok 4 Fast |
|---|---|---|
| LMArena Math | 1269 | 1419 |
| OTIS Mock AIME 2024-2025 | 1.1% | — |
| MATH Level 5 | 23% | — |
| GSM8K | 92% | — |
Knowledge Grok 4 Fast leads
GPT-4: 18.4 (#282), Grok 4 Fast: 32.0 (#214)
| Benchmark | GPT-4 | Grok 4 Fast |
|---|---|---|
| LMArena Expert | 1211 | 1411 |
| GPQA Diamond | 35.7% | — |
| Vectara Hallucination Rate | — | 19.7% |
| MMLU | 86.4% | — |
| TriviaQA | 84.8% | — |
Multilingual Grok 4 Fast leads
GPT-4: 40.6 (#215), Grok 4 Fast: 51.3 (#111)
| Benchmark | GPT-4 | Grok 4 Fast |
|---|---|---|
| LMArena Non-English | 1246 | 1396 |
| LMArena Chinese | 1242 | 1457 |
| LMArena French | 1283 | 1431 |
| LMArena German | 1251 | 1383 |
| LMArena Japanese | 1209 | 1352 |
| LMArena Korean | 1184 | 1357 |
| LMArena Russian | 1251 | 1389 |
| LMArena Spanish | 1261 | 1415 |
Instruction Following Grok 4 Fast leads
GPT-4: 65.3 (#222), Grok 4 Fast: 73.2 (#121)
| Benchmark | GPT-4 | Grok 4 Fast |
|---|---|---|
| LMArena Instruction Following | 1241 | 1387 |
Long Context Grok 4 Fast leads
GPT-4: 37.7 (#212), Grok 4 Fast: 63.2 (#3)
| Benchmark | GPT-4 | Grok 4 Fast |
|---|---|---|
| LMArena Longer Query | 1244 | 1415 |
| Fiction.LiveBench | — | 94.4% |
Writing & Preference Grok 4 Fast leads
GPT-4: 34.9 (#268), Grok 4 Fast: 60.0 (#102)
| Benchmark | GPT-4 | Grok 4 Fast |
|---|---|---|
| LMArena Text | 1263 | 1407 |
| LMArena Creative Writing | 1244 | 1387 |
| LMArena Multi-Turn | 1257 | 1414 |
| EQ-Bench Creative Writing | 752 | — |
Frequently asked questions
Is GPT-4 better than Grok 4 Fast?
Grok 4 Fast is the stronger model overall, scoring 39.4 to 29.1 on the Noometry Index.
Is GPT-4 or Grok 4 Fast better for coding?
They score almost the same on coding (31.6 vs 32.5); test both on your own repository before choosing.
How many benchmarks do GPT-4 and Grok 4 Fast share?
21 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Grok 4 Fast has 30.