News
SpaceXAI ships Grok 4.7 for coding and knowledge work
SpaceXAI’s Grok 4.7 lands in Cursor, Grok Build, and the API at Grok 4.6 prices, with longer-task training and a native Grok Bot harness.
Searcher → Analyst → Writer → Editor · subagentic-20260921-2000
SpaceXAI published Grok 4.7 on September 21, 2026 as its coding and knowledge-work flagship — a same-price step up from Grok 4.6 aimed at longer agent sessions.
The company says 4.7 uses a larger base model than 4.6 and a longer reinforcement-learning run on a harder mix of tasks, weighted toward work that takes many hours. SpaceXAI also trained the model to natively understand the Grok Bot harness, which it says helps on conversational tasks and general knowledge work. Developer docs list a 500,000-token context window, a May 2026 knowledge cutoff, text and image input with text output, and reasoning effort at low, medium, high (the default), or xhigh.
Pricing matches Grok 4.6: $2 per million input tokens and $6 per million output tokens. Grok 4.7 is live in Cursor, Grok Build (where it is the coding agent’s default), and the Grok API as grok-4.7. Docs also list OpenRouter, Vercel, and Cloudflare as model gateways, plus a US regional endpoint at https://us.api.x.ai/v1 with a 10% token-price premium.
A Fast variant runs on faster infrastructure at twice the standard token rates. It is available only in Cursor and Grok Build, is not on Grok Build’s free tier, and is not on the public xAI API.
On SpaceXAI’s own score table, Grok 4.7 posts 46.3% on CursorBench 4.0 versus 40.4% for Grok 4.6, and 38.0% on Terminal-Bench 4.0 versus 20.3%. Those figures are vendor-published, not independent evaluations. Cursor documents 4.7 as a first-party model with a 256k standard context window and 500k long context, four effort levels, and the same first-party credit treatment as 4.6. It is on every paid Cursor plan; Cursor Start locks it to medium effort without Fast mode. Composer stays available for everyday coding.
If you run coding agents, swap the model id to grok-4.7 in the Grok API, pick Grok 4.7 in Cursor, or try it in Grok Build — then read SpaceXAI’s model page for cache keys, context compaction, and reasoning effort.