SpaceXAI Grok 4.6 Launches at Parity With GPT-5.6 for Same Price

SpaceXAI released Grok 4.6, matching GPT- 5.6 benchmarks while keeping input tokens at $2 per million and output at $6 per million.

SpaceXAI Grok 4.6 Launches at Parity With GPT-5.6 for Same Price

SpaceXAI released Grok 4.6, an AI model update that aims to compete with the highest-tier industry benchmarks. This release matters because it positions the model at parity with GPT-5.6 Sol while maintaining the same pricing structure as its predecessor. Users and developers can now access frontier-level intelligence without a price increase, which directly impacts cost expectations for high-volume API usage.

SpaceXAI releases updated model matching GPT-5.6 benchmarks

The new model, Grok 4.6, represents a significant upgrade over Grok 4.5. SpaceXAI introduced the update on August 12, 2026, highlighting its performance capabilities across multiple independent benchmarks. The company claims the model delivers frontier intelligence and tops leadership boards on GDPVal-AA v2, AA-Briefcase, and Harvey LAB (Vals) benchmarks.

Under the hood, Grok 4.6 runs on massive compute infrastructure. The Colossus 1 data center supports over 220,000 GPUs, while the newer Colossus 2 facility boasts over 550,000 GPUs. These facilities include GB200 and GB300 accelerators, providing the raw power needed to achieve a score of 61 on the Artificial Analysis Intelligence Index.

Pricing for Grok 4.6 remains unchanged from the previous version. Input tokens cost $2 per 1 million tokens, and output tokens cost $6 per 1 million tokens. This pricing structure offers an 80 percent discount compared to the performance level of Fable 5, making it a cost-effective option for enterprises requiring high-throughput reasoning.

Discussion

0 comments

Log in to join the thread with a thoughtful take, question, or correction.

Add to the discussion