GPT-5 vs Grok 4
GPT-5 
Grok 4 Verdict
- GPT-5 and Grok 4 are level on quality (59.7 vs 59.8).
- GPT-5 is stronger in composite, human preference, knowledge, maths, multimodal, reasoning.
- Grok 4 is stronger in agents & tools, coding, instruction following, long context.
- Grok 4 is 2.2× cheaper ($1.56 vs $3.44 per 1M blended).
| Metric | GPT-5 | Grok 4 |
|---|---|---|
| BenchLeader Index | 59.7 | 59.8 |
| Agents & tools score | 50.0 | 53.4 |
| Coding score | 60.9 | 61.2 |
| Composite score | 52.8 | – |
| Human preference score | 61.7 | 59.8 |
| Instruction following score | 64.0 | 67.7 |
| Knowledge score | 60.8 | 56.0 |
| Long context score | 63.0 | 77.7 |
| Maths score | 72.8 | 58.0 |
| Multimodal score | 58.9 | 54.5 |
| Reasoning score | 64.2 | 58.3 |
| Blended price $/M | $3.44 | $1.56 |
| Output speed | 106 tok/s | – |
| Time to first answer | 54.4 s | – |
| Context window | 400k | 256k |
| GPQA Diamond | – | 87.0% |
| OTIS Mock AIME | – | 84.0% |
| Terminal-Bench | 49.6% | 27.2% |
| SimpleBench | – | 60.5% |
| Fiction.LiveBench 120k | – | 96.9% |
| SciCode | 42.9% | – |
| Cybench | – | 43.0% |
| Remote Labor Index | 1.7% | – |
| WeirdML | 39.8% | 45.7% |
| APEX-Agents | 18.3% | 15.2% |
| Epoch Capabilities Index | 150 | 146.4 |
| LMArena Text | 1427 | 1411 |
| LMArena Hard Prompts | 1449 | 1420 |
| LMArena Coding | 1462 | 1435 |
| LMArena Vision | 1232 | 1210 |
| AA Intelligence Index | 23.0 | – |
| IFBench | 73.1% | – |
| AA-LCR | 78.2% | – |
| MMMU-Pro | 74.2% | – |
| AA-Omniscience | -8.7 | – |
| Terminal-Bench Hard | 32.6% | – |
| GPQA Diamond (AA) | 85.3% | – |
| Humanity's Last Exam (AA) | 28.5% | – |
| τ²-Bench Telecom (AA) | 84.8% | – |
| AIME (Vals) | – | 90.6% |
| LiveCodeBench | – | 83.3% |
| MMLU-Pro | – | 85.3% |
| LegalBench | – | 83.2% |
| CorpFin | – | 66.0% |
| TaxEval | – | 65.1% |
| MedQA | – | 92.5% |
| MGSM | – | 90.9% |
| SWE-bench (Vals) | – | 57.8% |
| GPQA Diamond (Vals) | – | 88.1% |
| IMO 2025 | – | 21.4% |
| MathArena Apex | – | 2.1% |
| PRBench Finance | 51.3% | – |
| PRBench Legal | 49.0% | – |
| VISTA | 49.7% | – |
| MultiNRC | 52.1% | – |
| TutorBench | 55.3% | – |
| Kagi LLM Benchmark | 72.7% | 73.6% |
| IFEval (HELM) | 87.5% | 94.9% |
| Omni-MATH (HELM) | 64.7% | 60.3% |
| WildBench (HELM) | 85.7% | 79.7% |
| MMLU-Pro (HELM) | 86.3% | 85.1% |
| GPQA Diamond (HELM) | 79.1% | 72.6% |
| HELM Capabilities mean | 80.7% | 78.5% |
| Aider Polyglot | 88.0% | 79.6% |
| SWE-bench Verified (any scaffold) | 75.6% | – |
| ARC-AGI-1 | – | 79.6% |
| ARC-AGI-2 | – | 29.4% |
| BFCL Overall | – | 63.0% |
Data as of 2026-09-13. Best configuration of each model; every score links to its source on the model pages.
GPT-5 vs Grok 4: questions
- Is GPT-5 better than Grok 4?
- GPT-5 and Grok 4 are level on quality (59.7 vs 59.8). The BenchLeader Index combines every independent quality benchmark; Grok 4 is ahead overall as of 2026-09-13, but check the category scores for your use.
- Is GPT-5 better than Grok 4 for coding?
- Grok 4 scores higher in coding (61 vs 61 on the category index, where 50 is average).
- Is GPT-5 better than Grok 4 for agentic tasks?
- Grok 4 scores higher in agentic tasks (53 vs 50 on the category index, where 50 is average).
- Which is cheaper, GPT-5 or Grok 4?
- Grok 4 is cheaper: $1.56 against $3.44 per million tokens, blended at three input tokens per output token.
- Which has the larger context window?
- GPT-5 accepts more context: 400k against 256k tokens.