Breaking

SpaceXAI Releases Grok 4.7 With Coding and Agentic-Work Gains at Grok 4.6 Pricing

September 21, 2026 at 20:36 EDT

Foundation Models
AI Agents
Software Dev & Coding

SpaceXAI released Grok 4.7 on September 21, positioning it as its strongest model for coding, agentic tasks and professional knowledge work. The launch keeps the base API price at $2 per million input tokens and $6 per million output tokens while raising benchmark scores over Grok 4.6 and expanding distribution through developer tools.

September 21, 2026 - SpaceXAI

Grok 4.7 targets long-running AI work

SpaceXAI says the new model improves coding, agentic workflows and professional knowledge work while keeping Grok 4.6 base API pricing.

The cost watch: more output tokens

Output tokens per Intelligence Index task, shown as stacked blocks. One block is about 9k tokens.

27k

GPT-6 Astra Max

36k

Grok 4.6 High

81k

Grok 4.7 xHigh

Takeaway: the gains come with heavier reasoning-token usage, so per-task economics matter.

$2 / $6

input and output per 1M tokens

500k

token context window

Coding Agent Index score

3.3%

risky prompts allowed in HackerBench

How the model is positioned

Hard prompt and context

Longer reasoning loop

Tool use and self-checks

Coding or office deliverable

Why teams will test it

Higher coding-agent and AA-Briefcase scores
Same base API price as Grok 4.6
500k-token context for long work

What teams will measure

Output-token burn on real jobs
Retries, latency and human review
Cost per successful completion

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.