SpaceXAI launches Grok 4.7 for coding and knowledge work

SpaceXAI has announced Grok 4.7, presented on the company's news page as its most capable model for coding and knowledge work. The announcement, dated 21 September 2026, describes the model as “twice as fast, at half the price of comparable models”, and says it is served at the same price and speed as Grok 4.6.

What changed in the model

According to the announcement, Grok 4.7 uses a new, larger base model than Grok 4.6. It was trained with a longer reinforcement learning run on a harder mix of tasks, weighted toward problems that take many hours to complete. SpaceXAI says the model verifies its own work more carefully and manages longer context better, and that it was trained to natively understand the Grok Bot harness, which helps with conversational tasks and general knowledge work.

Benchmarks and pricing

The company's tables list Grok 4.7 (xHigh) at 46.3% on CursorBench 4.0, 71.0% on DeepSWE v1.1 — marked as a high-effort score — 64.0% on EEBench, 1,657 on AA Briefcase v1.1, 38.0% on Terminal-Bench 4.0, 19.6% on the Harvey Legal Agent Benchmark and 56.7% on HealthBench Professional. On the same tests, Grok 4.6 (High) is shown at 40.4%, 65.2%, 53.0%, 1,546, 20.3%, 15.8% and 48.5%. In the GDPval Elo chart, Grok 4.7 (xhigh) appears at 1,695, behind Fable 5.1 (max) at 1,735 and ahead of Grok 4.6 (high) at 1,605 and GPT-6 Astra (max) at 1,542. These are the vendor's own figures.

Pricing starts at $2 per million input tokens and $6 per million output tokens. A fast variant is offered with twice the output speed at twice the price. For comparison, the page lists GPT-5.6 Sol at $4/$20 and Fable 5.1 at $10/$50 per million tokens.

Safety and availability

SpaceXAI says Grok 4.7 was built with an entirely new safeguard stack and is the strongest model it has tested on refusals and jailbreak resistance. It reports that the model tops LatchBio's biosafety benchmark at 62.4% and, on HackerBench v0.3, allows only 3.3% of risky dual-use prompts through while rarely blocking legitimate security work. Selected cybersecurity partners have been given invite-only access to the model's red-team capabilities for defence research.

Grok 4.7 is available from launch in Cursor and Grok Build, and also through the Grok API, third-party coding harnesses, model routers and cloud platforms.