Gemini 3.1 Pro vs Grok 4
Gemini 3.1 Pro 
Grok 4 Verdict
- Gemini 3.1 Pro leads on quality: 62.8 vs 59.8.
- Gemini 3.1 Pro is stronger in agents & tools, composite, instruction following, knowledge, maths, multimodal, reasoning.
- Grok 4 is stronger in coding, long context.
- Grok 4 is 2.9× cheaper ($1.56 vs $4.50 per 1M blended).
| Metric | Gemini 3.1 Pro | Grok 4 |
|---|---|---|
| BenchLeader Index | 62.8 | 59.8 |
| Agents & tools score | 62.6 | 53.4 |
| Coding score | 59.8 | 61.2 |
| Composite score | 61.5 | – |
| Human preference score | 59.8 | 59.8 |
| Instruction following score | 68.4 | 67.7 |
| Knowledge score | 64.9 | 56.0 |
| Long context score | 65.2 | 77.7 |
| Maths score | 60.7 | 58.0 |
| Multimodal score | 65.8 | 54.5 |
| Reasoning score | 69.8 | 58.3 |
| Blended price $/M | $4.50 | $1.56 |
| Output speed | 115 tok/s | – |
| Time to first answer | 23.4 s | – |
| Context window | 1.0M | 256k |
| GPQA Diamond | 94.1% | 87.0% |
| FrontierMath Tiers 1–3 | 59.6% | – |
| FrontierMath Tier 4 | 26.8% | – |
| OTIS Mock AIME | 95.6% | 84.0% |
| Humanity's Last Exam | 46.4% | – |
| Terminal-Bench | 80.2% | 27.2% |
| SimpleBench | 79.6% | 60.5% |
| Fiction.LiveBench 120k | – | 96.9% |
| SciCode | 58.9% | – |
| Cybench | – | 43.0% |
| WeirdML | 72.1% | 45.7% |
| APEX-Agents | 33.5% | 15.2% |
| ProofBench | 26.0% | – |
| GSO-Bench | 22.6% | – |
| Epoch Capabilities Index | 155 | 146.4 |
| LMArena Text | 1487 | 1411 |
| LMArena Hard Prompts | 1508 | 1420 |
| LMArena Coding | 1520 | 1435 |
| LMArena WebDev | 1447 | – |
| LMArena Vision | 1295 | 1210 |
| LMArena Agent | -5.6 | – |
| AA Intelligence Index | 30.4 | – |
| IFBench | 77.1% | – |
| AA-LCR | 82.0% | – |
| MMMU-Pro | 82.4% | – |
| AA-Omniscience | 31.9 | – |
| Terminal-Bench Hard | 53.8% | – |
| GPQA Diamond (AA) | 94.1% | – |
| Humanity's Last Exam (AA) | 47.0% | – |
| SciCode (AA) | 58.7% | – |
| τ²-Bench Telecom (AA) | 95.6% | – |
| 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% |
| AIME 2026 | 98.3% | – |
| HMMT February 2026 | 94.7% | – |
| IMO 2025 | – | 21.4% |
| MathArena Apex | 60.9% | 2.1% |
| MultiChallenge | 71.4% | – |
| PRBench Finance | 41.9% | – |
| PRBench Legal | 44.0% | – |
| MultiNRC | 64.7% | – |
| HiL-Bench | 35.3% | – |
| TutorBench | 53.0% | – |
| EQ-Bench 4 | 1142 | – |
| Kagi LLM Benchmark | – | 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% |
| ARC-AGI-1 | 98.0% | 79.6% |
| ARC-AGI-2 | 77.1% | 29.4% |
| ARC-AGI-3 | 0.4% | – |
| 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.
Gemini 3.1 Pro vs Grok 4: questions
- Is Gemini 3.1 Pro better than Grok 4?
- Gemini 3.1 Pro leads on quality: 62.8 vs 59.8. The BenchLeader Index combines every independent quality benchmark; Gemini 3.1 Pro is ahead overall as of 2026-09-13, but check the category scores for your use.
- Is Gemini 3.1 Pro better than Grok 4 for coding?
- Grok 4 scores higher in coding (61 vs 60 on the category index, where 50 is average).
- Is Gemini 3.1 Pro better than Grok 4 for agentic tasks?
- Gemini 3.1 Pro scores higher in agentic tasks (63 vs 53 on the category index, where 50 is average).
- Which is cheaper, Gemini 3.1 Pro or Grok 4?
- Grok 4 is cheaper: $1.56 against $4.50 per million tokens, blended at three input tokens per output token.
- Which has the larger context window?
- Gemini 3.1 Pro accepts more context: 1.0M against 256k tokens.