Grok 4.7 API: xAI's new flagship on FastMetal
Grok 4.7 is live on FastMetal as grok-4.7. It replaces Grok 4.6 as the top model from SpaceXAI (formerly xAI), and it is aimed at coding and knowledge work.
What changed from Grok 4.6
SpaceXAI describes Grok 4.7 as a model that stays on hard tasks longer and checks its own work more carefully. It runs at the same speed as Grok 4.6, on a larger base model with more reinforcement learning on long tasks. Those are the vendor's claims; we have not benchmarked them.
What we did check, with real requests through our gateway on launch day:
- Image input: the model described an image we sent.
- Tool calling: it called a tool when one was offered.
- Structured output: it returned JSON that matched a strict JSON Schema.
The context window is 500,000 tokens, enough to pass a long document or a sizable slice of a codebase in one request.
Pricing
Grok 4.7 is billed per million input tokens and per million output tokens, with a higher tier for prompts above 200,000 tokens. At launch it cost less than Grok 4.6. SpaceXAI raised its list prices on September 28, 2026, and the two models now bill at the same rates. Current rates are on the Grok 4.7 model page.
Calling it
The endpoint is OpenAI-compatible, so existing code only needs a new model value:
curl https://api.fastmetal.ai/v1/chat/completions \
-H "Authorization: Bearer $FASTMETAL_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "grok-4.7",
"messages": [{"role": "user", "content": "Hello"}]
}'
Your existing API key and balance work as they are. To see the two generations side by side, open the Grok 4.6 vs Grok 4.7 comparison. Everything else we serve is in the model catalog.