Grok 4.5 vs Grok 4.6
| model Grok 4.5 xAI | model Grok 4.6 xAI | |
|---|---|---|
| Context | 500K | 500K |
| Input $/M | — | — |
| Output $/M | — | — |
| Knowledge cutoff | — | 2026-02 |
| Available via | xAI API | xAI API |
| SWE-bench Pro | 64.7% | — |
| SWE-bench Multilingual | 78% | — |
| Terminal-Bench (success rate) | 83.3 | — |
| DeepSWE | 62% | — |
| SWE Marathon (pass@1) | 29 | — |