Model
Model explorer
Grok 4.7
CLOSEDSpaceXAI · Grok 4.7 family · released Sep 21, 2026
SpaceXAI's Sep 21, 2026 flagship, available the same day on the xAI API as grok-4.7, in Cursor and Grok Build, at Grok 4.6's unchanged $2/$6 per Mtok with a 500K context. The launch post says it uses a new, larger base model and a longer RL run weighted to multi-hour tasks. At xHigh effort its launch-table rows are CursorBench 4.0 46.3%, DeepSWE v1.1 71.0% (high effort), AA Briefcase v1.1 1,657, Terminal-Bench 4.0 37.6% and Harvey Legal Agent 19.6%, well ahead of Grok 4.6 High on every row but behind Fable 5.1 Max on most coding rows.
ReasoningCodingVisionFunction callingTool useAgentic
2534.9
Elo · rank #36
Parameters
Undisclosed
Active params
Undisclosed
Context
500K tokens
Architecture
Proprietary sparse-MoE reasoning model (undisclosed size); 500K-token context
License
Proprietary
Languages
—
API price (in/out)
$2 / $6
Modalities
text · vision
Benchmark results
Bar shows position within the tracked field; marker = field best
best: Claude Fable 5 · 98.5%
best: GPT-6 Astra · 95.0%
best: GPT-6 Astra · 62.7%
best: Claude Fable 5 · 1504
best: Claude Opus 5.5 · 57.8%
best: Gemini 4 Argon · 77.9%
best: Gemini 4 Argon · 19.6%
best: Claude Opus 5.5 · 65.6%
best: Claude Sonnet 5.5 · 70.6%
Run it locally
Closed weights — available via API only. No local deployment.
Input / M tok
$2
Output / M tok
$6
API price $2/$6 · each benchmark row carries its own source badge (see methodology)