GPT-5.2 vs Grok 4
GPT-5.2 
Grok 4 Verdict
- GPT-5.2 and Grok 4 are level on quality (60.3 vs 59.8).
- GPT-5.2 is stronger in agents & tools, composite, human preference, knowledge, multimodal, reasoning.
- Grok 4 is stronger in coding, instruction following, long context, maths.
- Grok 4 is 3.1× cheaper ($1.56 vs $4.81 per 1M blended).
| Metric | GPT-5.2 | Grok 4 |
|---|---|---|
| BenchLeader Index | 60.3 | 59.8 |
| Agents & tools score | 60.4 | 53.4 |
| Coding score | 53.7 | 61.2 |
| Composite score | 61.6 | – |
| Human preference score | 67.8 | 59.8 |
| Instruction following score | 65.4 | 67.7 |
| Knowledge score | 58.8 | 56.0 |
| Long context score | 65.6 | 77.7 |
| Multimodal score | 59.9 | 54.5 |
| Reasoning score | 61.3 | 58.3 |
| Maths score | – | 58.0 |
| Blended price $/M | $4.81 | $1.56 |
| Output speed | 72 tok/s | – |
| Time to first answer | 90.8 s | – |
| Context window | 400k | 256k |
| GPQA Diamond | – | 87.0% |
| OTIS Mock AIME | – | 84.0% |
| Humanity's Last Exam | 27.8% | – |
| Terminal-Bench | 64.9% | 27.2% |
| SimpleBench | 45.8% | 60.5% |
| Fiction.LiveBench 120k | – | 96.9% |
| Cybench | – | 43.0% |
| Remote Labor Index | 2.1% | – |
| WeirdML | – | 45.7% |
| APEX-Agents | 23.0% | 15.2% |
| Epoch Capabilities Index | 153.5 | 146.4 |
| LMArena Text | 1476 | 1411 |
| LMArena Hard Prompts | 1497 | 1420 |
| LMArena Coding | 1515 | 1435 |
| LMArena WebDev | 1417 | – |
| LMArena Vision | 1268 | 1210 |
| AA Intelligence Index | 30.4 | – |
| IFBench | 75.4% | – |
| AA-LCR | 82.7% | – |
| AA-Omniscience | -0.9 | – |
| Terminal-Bench Hard | 47.0% | – |
| GPQA Diamond (AA) | 90.3% | – |
| Humanity's Last Exam (AA) | 37.7% | – |
| 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% |
| SWE-Bench Pro | 29.9% | – |
| VISTA | 46.6% | – |
| MultiNRC | 42.2% | – |
| TutorBench | 53.5% | – |
| Kagi LLM Benchmark | 73.3% | 73.6% |
| IFEval (HELM) | – | 94.9% |
| Omni-MATH (HELM) | – | 60.3% |
| WildBench (HELM) | – | 79.7% |
| MMLU-Pro (HELM) | – | 85.1% |
| GPQA Diamond (HELM) | – | 72.6% |
| HELM Capabilities mean | – | 78.5% |
| Aider Polyglot | – | 79.6% |
| SWE-bench Verified (bash only) | 69.0% | – |
| ARC-AGI-1 | 94.5% | 79.6% |
| ARC-AGI-2 | 72.9% | 29.4% |
| BFCL Overall | 55.9% | 63.0% |
Data as of 2026-09-13. Best configuration of each model; every score links to its source on the model pages.
GPT-5.2 vs Grok 4: questions
- Is GPT-5.2 better than Grok 4?
- GPT-5.2 and Grok 4 are level on quality (60.3 vs 59.8). The BenchLeader Index combines every independent quality benchmark; GPT-5.2 is ahead overall as of 2026-09-13, but check the category scores for your use.
- Is GPT-5.2 better than Grok 4 for coding?
- Grok 4 scores higher in coding (61 vs 54 on the category index, where 50 is average).
- Is GPT-5.2 better than Grok 4 for agentic tasks?
- GPT-5.2 scores higher in agentic tasks (60 vs 53 on the category index, where 50 is average).
- Which is cheaper, GPT-5.2 or Grok 4?
- Grok 4 is cheaper: $1.56 against $4.81 per million tokens, blended at three input tokens per output token.
- Which has the larger context window?
- GPT-5.2 accepts more context: 400k against 256k tokens.