Grok 4.7

Same price as 4.6; longer-horizon agentic work.

Released

Sep 21, 2026

Price in / out

$2 / $6

Blended $/1M

Context

500K

Benchmark scoresvs Grok 4.6

Benchmark	Score	Δ vs prior	Rank (current)	Configuration	Source
Terminal-Bench 2.1	73.4%	-4.9	#9 of 20	leaderboard	benchlm.ai ↗

Grok versions

Timeline

Version	Released	$ in / out	Terminal-Bench	SWE-bench	GPQA	HLE	ARC-AGI-2	τ²-bench
Grok 4Superseded	2025-07-09	$3 / $15	—	—	87.5%	25.4%	15.9%	74.9%
Grok 4.1Superseded	2025-11-17	$3 / $15	—	—	—	—	—	—
Grok 4.20Superseded	2026-03-10	$1.25 / $2.5	44.2%	76.7%	88.5%	—	53.3%	—
Grok 4.3Superseded	2026-05-01	$1.25 / $2.5	41.9%	—	—	35%	—	97.7%
Grok 4.5Superseded	2026-07-08	$2 / $6	67.8%	—	—	—	52.6%	—
Grok 4.6Superseded	2026-08-12	$2 / $6	78.3%	—	—	—	67.1%	—
Grok 4.7	2026-09-21	$2 / $6	73.4%	—	—	—	—	—

Compare

Open the Comparison Lab with Grok 4.7 and its closed-weight peers pre-selected, then toggle any benchmark or model.

Every Grok version over time

Shipping with it?

Benchmarks are a starting point. If you run Grok 4.7 in production, post what it does well and where it falls over; other founders are deciding right now.

Write about Grok 4.7

Current closed-weight peers

All

Claude Sonnet 5.5$2 in Claude Opus 5.5$4 in Claude Fable 5.1$10 in GPT-6 Luna$0.1 in GPT-6 Astra$10 in Gemini 3.8 Flash$0.75 in