---
format: "aidr-story-markdown/v1"
id: "5a022d30397b690537174689616e4ffe65700ccef2d887a4562258b6ebd7c318"
canonical_url: "https://aidr.today/5a022d30?lang=en"
title: "SpaceXAI Launches Grok 4.7 With 19.6% Legal Benchmark Score to Beat GPT 5.6"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-09-22T03:11:03.000Z"
category: "Models"
topics: ["spacexai","grok","regulation","benchmark","api"]
source_urls: ["https://huggingnews.com/ai/spacexai-launches-grok-47-with-196percent-legal-benchmark-score-to-beat-3a4eee99","https://x.com/elonmusk/status/2102102621037236699","https://x.com/rohanpaul_ai/status/2102107446852592045","https://x.com/theo/status/2102209324491891005","https://x.com/davis7/status/2102213709825560595","https://x.com/SpaceXAI/status/2102069815225586149","https://x.com/mark_k/status/2102071475414986981","https://x.com/scaling01/status/2102067894863409493"]
summary: "The new AI model is available via Grok Build, Cursor, and API for developers specializing in coding and knowledge work. SpaceXAI's latest release scored 19.6% on the Harvey Legal Agent Benchmark, outperforming GPT-5.6 Sol Max at 2.5% and Fable 5.1 Max at 6.7%. The model also reached 38.0% on Terminal-Bench 4.0 compared to 20.3% for Grok 4.6, while headline pricing remains at $2 per 1 million input tokens and $6 per 1 million output tokens. Early user reports indicate pricing volatility and restrictive usage limits on the $300 per month supergrok heavy plan. Costs double for tasks exceeding 200,000 tokens or when utilizing a high-speed mode, according to some users. Elon Musk recommended utilizing the company's Build harness to achieve optimal results with the model's updated reasoning and self-verification capabilities."
---

# SpaceXAI Launches Grok 4\.7 With 19\.6% Legal Benchmark Score to Beat GPT 5\.6

> [Open the canonical story](<https://aidr.today/5a022d30?lang=en>)

**Published:** 2026-09-22T03:11:03.000Z
**Category:** Models
**Topics:** spacexai, grok, regulation, benchmark, api

## Summary

The new AI model is available via Grok Build, Cursor, and API for developers specializing in coding and knowledge work\. SpaceXAI's latest release scored 19\.6% on the Harvey Legal Agent Benchmark, outperforming GPT\-5\.6 Sol Max at 2\.5% and Fable 5\.1 Max at 6\.7%\. The model also reached 38\.0% on Terminal\-Bench 4\.0 compared to 20\.3% for Grok 4\.6, while headline pricing remains at $2 per 1 million input tokens and $6 per 1 million output tokens\. Early user reports indicate pricing volatility and restrictive usage limits on the $300 per month supergrok heavy plan\. Costs double for tasks exceeding 200,000 tokens or when utilizing a high\-speed mode, according to some users\. Elon Musk recommended utilizing the company's Build harness to achieve optimal results with the model's updated reasoning and self\-verification capabilities\.

## Sources

- [Story source](<https://huggingnews.com/ai/spacexai-launches-grok-47-with-196percent-legal-benchmark-score-to-beat-3a4eee99>)
- [Story source](<https://x.com/elonmusk/status/2102102621037236699>)
- [Supporting source](<https://x.com/rohanpaul_ai/status/2102107446852592045>)
- [Supporting source](<https://x.com/theo/status/2102209324491891005>)
- [Supporting source](<https://x.com/davis7/status/2102213709825560595>)
- [Story source](<https://x.com/SpaceXAI/status/2102069815225586149>)
- [Supporting source](<https://x.com/mark_k/status/2102071475414986981>)
- [Supporting source](<https://x.com/scaling01/status/2102067894863409493>)