Grok 4.7
Grok 4.7 is xAI's most capable model for coding and knowledge work, released September 21, 2026 with the API ID grok-4.7.
Provider
SpaceXAI
Model family
xAI Grok
Frontier multimodal agentic LLM
Cost tier
Grok
Status
Current
Release Sep 21, 2026
Technical specs
- Inputs
- text, image
- Outputs
- text
- License
- Proprietary API
- API ID
grok-4.7- Context tokens
- 500,000
- Reasoning Levels
- low, medium, high, xhigh
- Input Usd Per Million
- 2
- Output Usd Per Million
- 6
- Default Reasoning Level
- high
- Cache Read Usd Per Million
- 0.5
- Long Context Threshold Tokens
- 200,000
- Long Context Input Usd Per Million
- 4
- Long Context Output Usd Per Million
- 12
- Long Context Cache Read Usd Per Million
- 1
- Capabilities
- 500,000-token context window
- Reasoning levels low, medium, high, and xhigh
- Native understanding of the Grok Bot harness
- Available in Cursor, Grok Build, the Grok API, and cloud platforms
Why teams choose it
Same list price as Grok 4.6, so most Grok 4.6 workloads can switch IDs without a cost change.
Default reasoning level is high; xhigh is the setting behind the headline CursorBench score.
Tradeoffs to know
- Prompts over 200,000 tokens bill at double the rate.
- Grok 4.7 Fast is only offered in Cursor and Grok Build.
- Benchmark figures on the launch post are vendor-reported.
Benchmarks
- Eebench Percent (vendor-reported)
- 64
- Aa Briefcase V1 1 (vendor-reported)
- 1,657
- Cursorbench 4 0 Percent (vendor-reported)
- 46.3
- Deepswe V1 1 Percent High (vendor-reported)
- 71
- Terminal Bench 4 0 Percent (vendor-reported)
- 37.6
- Healthbench Professional Percent (vendor-reported)
- 56.7
xAI Grok family lineup
Previous versions
Related models
SpaceXAI
Grok 1.5
Current · 1.5 · grok
SpaceXAI
Grok Voice 3.0
Grok Voice 3.0 is xAI's speech-to-speech voice model, released October 14, 2026 with the API ID grok-voice-3.0 (alias…
SpaceXAI
Grok Imagine Video 1.5
Grok Imagine Video 1.5 is xAI's full-quality video generation model on the Grok API. xAI's video docs list native 108…
Grok 4.7 FAQ
What is Grok 4.7?
7. 6. It takes text and image input and returns text, with a 500,000-token context window and reasoning levels low, medium, high (the default), and xhigh. 50 cached input, and $6 output per million tokens; above 200,000 it costs $4, $1, and $12. 1 at high effort. A fast variant at twice the output speed and twice the price is offered in Cursor and Grok Build.
When does Grok 4.7 fit best?
Long-running coding agents in Cursor and Grok Build
What should teams watch out for with Grok 4.7?
Prompts over 200,000 tokens bill at double the rate.
Explore next
Models, tools, and comparisons that connect to this reference.