Grok 4.6 vs o3
Grok 4.6 
o3 Verdict
- Grok 4.6 (medium) leads on quality: 63.9 vs 61.6.
- Grok 4.6 (medium) is stronger in coding, composite, knowledge, long context, reasoning.
- o3 is stronger in agents & tools, human preference, instruction following, maths, multimodal.
- Grok 4.6 (medium) is 1.2× cheaper ($3.00 vs $3.50 per 1M blended).
- o3 streams 2.1× faster (107 vs 52 tokens per second).
| Metric | Grok 4.6 (medium) | o3 |
|---|---|---|
| BenchLeader Index | 63.9 | 61.6 |
| Coding score | 64.1 | 62.0 |
| Composite score | 83.5 | 54.5 |
| Knowledge score | 77.6 | 58.0 |
| Long context score | 67.1 | 63.8 |
| Reasoning score | 62.1 | 61.5 |
| Agents & tools score | – | 70.3 |
| Human preference score | – | 62.3 |
| Instruction following score | – | 65.0 |
| Maths score | – | 78.6 |
| Multimodal score | – | 57.9 |
| Blended price $/M | $3.00 | $3.50 |
| Output speed | 52 tok/s | 107 tok/s |
| Time to first answer | 26.1 s | 6.3 s |
| Context window | 500k | 200k |
| SciCode | 54.6% | – |
| Epoch Capabilities Index | – | 146.9 |
| LMArena Text | – | 1432 |
| LMArena Hard Prompts | – | 1441 |
| LMArena Coding | – | 1460 |
| LMArena Vision | – | 1214 |
| AA Intelligence Index | 43.0 | 20.2 |
| IFBench | – | 71.4% |
| AA-LCR | 81.0% | 74.7% |
| MMMU-Pro | – | 70.1% |
| AA-Omniscience | 28 | -15.6 |
| Terminal-Bench Hard | – | 37.1% |
| GPQA Diamond (AA) | 93.5% | 82.7% |
| Humanity's Last Exam (AA) | 42.1% | 20.1% |
| SciCode (AA) | 55.9% | – |
| τ²-Bench Telecom (AA) | – | 80.7% |
| PRBench Finance | – | 47.7% |
| PRBench Legal | – | 48.6% |
| MMMU (validation) | – | 82.9% |
| MMMU-Pro (official) | – | 76.4% |
| Kagi LLM Benchmark | – | 67.6% |
| IFEval (HELM) | – | 86.9% |
| Omni-MATH (HELM) | – | 71.4% |
| WildBench (HELM) | – | 86.1% |
| MMLU-Pro (HELM) | – | 85.9% |
| GPQA Diamond (HELM) | – | 75.3% |
| HELM Capabilities mean | – | 81.1% |
| Aider Polyglot | – | 81.3% |
| SWE-bench Verified (bash only) | – | 58.4% |
| SWE-bench Verified (any scaffold) | – | 58.4% |
| ARC-AGI-1 | 87.5% | – |
| ARC-AGI-2 | 61.3% | – |
| BFCL Overall | – | 63.0% |
Data as of 2026-09-12. Best configuration of each model; every score links to its source on the model pages.
Grok 4.6 vs o3: questions
- Is Grok 4.6 better than o3?
- Grok 4.6 (medium) leads on quality: 63.9 vs 61.6. The BenchLeader Index combines every independent quality benchmark; Grok 4.6 (medium) is ahead overall as of 2026-09-12, but check the category scores for your use.
- Is Grok 4.6 better than o3 for coding?
- Grok 4.6 scores higher in coding (64 vs 62 on the category index, where 50 is average).
- Which is cheaper, Grok 4.6 or o3?
- Grok 4.6 is cheaper: $3.00 against $3.50 per million tokens, blended at three input tokens per output token.
- Which is faster, Grok 4.6 or o3?
- o3 streams faster: 107 against 52 output tokens per second.
- Which has the larger context window?
- Grok 4.6 accepts more context: 500k against 200k tokens.