x.ai released Grok 4.7 as its most capable model for coding and knowledge work. It uses a new, larger base model trained on harder, longer tasks. It verifies its own work better and manages longer context. It natively understands the Grok Bot harness for conversational tasks. It leads on price-performance in CursorBench 4.0. It tops LatchBio’s biosafety benchmark at 62.4%. It allows only 3.3% of risky dual-use prompts through on HackerBench v0.3. It is priced at $2 per million input tokens and $6 per million output tokens. A fast variant offers twice the output speed at twice the price. It is available today in Cursor, Grok Build, and the Grok API. Select cybersecurity partners get invite-only access to its red-team capabilities.
What changed
Grok 4.6 was the prior most capable model for coding and knowledge work.
What it unlocks
Running longer tasks, verifying its own work, and managing longer context more effectively.
- $2 per million input tokens
- $6 per million output tokens
- 62.4% on LatchBio biosafety benchmark
- 3.3% risky dual-use prompts allowed on HackerBench v0.3
What you need to act on it
- access via Cursor, Grok Build, Grok API, or third-party platforms
Sources