Grok 4.3 vs Grok 4.6
Grok 4.3 Verdict
- Grok 4.6 (medium) leads on quality: 63.9 vs 60.7.
- Grok 4.3 (medium) is stronger in agents & tools, instruction following, multimodal.
- Grok 4.6 (medium) is stronger in composite, knowledge, long context, coding, reasoning.
- Grok 4.3 (medium) is 1.9× cheaper ($1.56 vs $3.00 per 1M blended).
- Grok 4.3 (medium) streams 2.1× faster (112 vs 53 tokens per second).
| Metric | Grok 4.3 (medium) | Grok 4.6 (medium) |
|---|---|---|
| BenchLeader Index | 60.7 | 63.9 |
| Agents & tools score | 60.1 | – |
| Composite score | 60.3 | 83.4 |
| Instruction following score | 80.0 | – |
| Knowledge score | 72.1 | 77.5 |
| Long context score | 64.0 | 67.1 |
| Multimodal score | 60.2 | – |
| Coding score | – | 64.1 |
| Reasoning score | – | 62.1 |
| Blended price $/M | $1.56 | $3.00 |
| Output speed | 112 tok/s | 53 tok/s |
| Time to first answer | 12.1 s | 33.4 s |
| Context window | 1M | 500k |
| SciCode | – | 54.6% |
| AA Intelligence Index | 24.8 | 43.0 |
| IFBench | 83.3% | – |
| AA-LCR | 75.0% | 81.0% |
| MMMU-Pro | 75.8% | – |
| AA-Omniscience | 16.7 | 28 |
| Terminal-Bench Hard | 30.3% | – |
| GPQA Diamond (AA) | 89.0% | 93.5% |
| Humanity's Last Exam (AA) | 30.0% | 42.1% |
| SciCode (AA) | – | 55.9% |
| τ²-Bench Telecom (AA) | 91.2% | – |
| ARC-AGI-1 | – | 87.5% |
| ARC-AGI-2 | – | 61.3% |
Data as of 2026-09-10. Best configuration of each model; every score links to its source on the model pages.
Grok 4.3 vs Grok 4.6: questions
- Is Grok 4.3 better than Grok 4.6?
- Grok 4.6 (medium) leads on quality: 63.9 vs 60.7. The BenchLeader Index combines every independent quality benchmark; Grok 4.6 (medium) is ahead overall as of 2026-09-10, but check the category scores for your use.
- Which is cheaper, Grok 4.3 or Grok 4.6?
- Grok 4.3 is cheaper: $1.56 against $3.00 per million tokens, blended at three input tokens per output token.
- Which is faster, Grok 4.3 or Grok 4.6?
- Grok 4.3 streams faster: 112 against 53 output tokens per second.
- Which has the larger context window?
- Grok 4.3 accepts more context: 1M against 500k tokens.