Model · SpaceXAI · Text
Grok 4.6
The frontier model from SpaceXAI (formerly xAI), built for long-running agents and ambitious interactive work. SpaceXAI's docs put it level with GPT-5.6 Sol on the Artificial Analysis Intelligence Index (2026-09-08) at a fraction of the price.
- Modality
- Text
- License
- Proprietary (Proprietary)
- Context window
- 500,000 tokens 500k tokens. There is no published max-output ceiling — the API's max_output_tokens defaults to 128,000 when unset and can be raised. Prompts that reach 200k tokens are billed at the higher tier for every token in the request, including the ones below the threshold.
- Released
- August 12, 2026
- Last verified
- September 8, 2026
- Runs locally
- No
- Also handles
- vision, reasoning, code
Strengths
- Frontier-level score — 61 on the Artificial Analysis Intelligence Index as reported by SpaceXAI's own docs on 2026-09-08, level with GPT-5.6 Sol — at $2/$6 per MTok
- Built to stay with complex tasks across many steps — research, codebase work, turning an idea into a working application
- More self-testing and verification on long runs than Grok 4.5
- Cached input at $0.50 per MTok, a quarter of the standard input price
Weaknesses
- Reasoning cannot be disabled — short, simple calls still pay for thinking tokens
- The 200k-token price tier is a cliff, not a gradient: crossing it doubles the cost of the whole request
- No Batch API support
- February 2026 knowledge cutoff — older than most of its frontier peers
- Closed weights — API only
Try it
| Where | Type | Notes |
|---|---|---|
| xAI API | hosted-api | API key required; us-east-1 and us-west-2 |
Used in solutions
Official sources
- Model docs docs
- Announcement announcement
- Pricing docs
Change log
- — Initial entry — first SpaceXAI model in the catalog. Released 2026-08-12. Maker recorded as SpaceXAI: that is the string on the copyright notice, the nav on every announcement page, the model docs' own prose, and the Arena leaderboard's provider column. The company was formerly xAI, and the API and SDK identifiers are still xai_sdk / XAI_API_KEY / api.x.ai. Reasoning controls run low / medium / high / xhigh with no way to disable.
Esc