| Date | 21 Sep 2026 |
|---|
xAI released Grok 4.7 on September 21, 2026 as its most capable model for coding and knowledge work, delivering measurable gains on multi-hour tasks at the same $2/$6 pricing as Grok 4.6 while introducing a new safeguard stack.
What Actually Changed
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. The model improves at verifying its own work and managing longer context. It was also trained to natively understand the Grok Bot harness.
Pricing remains $2 per million input tokens and $6 per million output tokens. A fast variant delivers twice the output speed at twice the price. Context window is 500,000 tokens. Reasoning effort levels are low, medium, high (default), and xhigh.
Availability is immediate in Cursor, Grok Build, the Grok API, third-party coding harnesses, model routers, and cloud platforms.
xAI-reported benchmark results (Grok 4.7 at xHigh or high effort vs. prior and selected peers):
| Benchmark | Grok 4.7 | Grok 4.6 | GPT-5.6 Sol Max | Fable 5.1 Max |
|---|---|---|---|---|
| CursorBench 4.0 | 46.3% | 40.4% | 41.7% | 51.8% |
| DeepSWE v1.1 | 71.0%* | 65.2% | 72.7% | 70.0% |
| Terminal-Bench 4.0 | 38.0% | 20.3% | 37.3% | 57.9% |
| EEBench | 64.0% | 53.0% | 39.4% | 56.4% |
| Harvey Legal Agent Benchmark | 19.6% | 15.8% | 2.5% | 6.7% |
| HealthBench Professional | 56.7% | 48.5% | 60.5% | 62.1% |
| AA Briefcase v1.1 | 1,657 | 1,546 | 1,487 | 1,678 |
*High-effort score. Grok 4.7 also improves on GDPval relative to Grok 4.6 and performs comparably to other frontier models on professional knowledge-work tasks.
Safety: Grok 4.7 ships with an entirely new safeguard stack. xAI states it is the strongest model the company has tested on refusals and jailbreak resistance. It leads on dual-use utility and safe refusal in cybersecurity and biological domains, scoring 62.4% on LatchBio’s biosafety benchmark. On HackerBench v0.3 it allows only 3.3% of risky dual-use prompts through while rarely blocking legitimate security work. Select cybersecurity partners have been given invite-only access to its red-team capabilities.
Signal Implications
- Infrastructure & product reality: The combination of longer-horizon gains, unchanged low pricing, and broad availability in Cursor and the Grok API strengthens Grok’s position as a high-volume coding and agent workhorse rather than a pure frontier claim.
- Product continuity / vendor risk: Same price and speed as Grok 4.6 plus immediate multi-platform access lowers switching friction for existing users while the new safeguard stack adds a concrete safety data point for diligence.
- Rights, regulatory, or diligence surface: Explicit dual-use safety metrics and partner red-team access provide primary-source material for enterprise risk reviews and buyer confidence conversations.
- Capital & buyer confidence: Sustained aggressive pricing alongside measurable multi-hour task improvement reinforces a cost-performance narrative that sophisticated buyers and investors track when evaluating category economics.
Who Needs to Act / Check
Founders and technical leads running coding agents or multi-hour knowledge workflows should test Grok 4.7 in Cursor or via the API against their own repositories and harnesses, paying attention to effort level and token usage. Communications advisors preparing investor or enterprise narratives can cite the primary pricing, benchmark table, and safeguard claims directly. Teams already on Grok 4.6 face near-zero migration cost.