Artificial Analysis Intelligence found that Grok 4.6, SpaceXAI's new model, tied with OpenAI's GPT-5.6 Sol by scoring 61 points, and is more than 60% cheaper than its main competitors.
In the ranking, only Claude Opus 5, with 63 points, and Claude Fable 5, with 62, both from Anthropic, are ahead. The new model's score represents a five-point jump over its predecessor, Grok 4.5.
Performance on agentic tasks
On the GDPval-AA v2 benchmark, which evaluates real knowledge work performed on a computer, Grok 4.6 ranks second, with an Elo of 1,753, behind only Claude Opus 5. SpaceXAI's model completes complex tasks in about 53 steps, while Claude Opus 5 takes about 103.
Price and availability
The price of Grok 4.6 remains at US$ 2 and US$ 6 per million tokens, more than 60% lower than the price of Claude Opus 5 (US$ 5 and US$ 25) and that of GPT-5.6 Sol (US$ 5 and US$ 30).
The model is available via API, Cursor, Grok Build, and partners such as OpenRouter, Vercel, and Cloudflare. During the first week, x.ai is offering double the usage quota on Grok Build and Cursor.
