xAIGA
Grok 4.6
Current xAI flagship: same-scale agent refresh of 4.5, AA Index 61, $2/$6 under 200k. No Starlight arena receipt yet.
Read the full Grok 4.6 analysisContext
500K
Max output
—
Input /1M
$2.00
Output /1M
$6.00
Live pricing via OpenRouter
Best for
- Long-running coding and knowledge-work agents on the xAI API
- First-pass app and visual structure work (vendor-stated focus)
- Frontier-priced loops that must stay under the 200k prompt cliff
Watch out
Not a documented new base model. Cached input rose to $0.50. Price doubles past 200k prompt tokens. Do not treat vendor or AA tables as SIS Model Arena winners.
For creators. Use as the default Grok API model for multi-step build loops; keep Claude on receipt-gated SIS arena work until a Grok 4.6 JSON receipt exists.
Benchmarks
| aa intelligence index | 61 |
| gdpval aa | 1753 |
| aa briefcase | 1577 |
| apex agents | 57.5 |
| deepswe v1 1 | 65.9 |
| terminal bench v3 | 26 |
| cursorbench v3 2 | 69.9 |
Capabilities
- Long-running agent stamina (vendor-stated focus)
- 500k context window
- Configurable reasoning including xhigh
- Text + image input, text output
- Unchanged $2/$6 list price under 200k prompt tokens