Grok 4.6 delivers frontier intelligence at 60%+ lower cost than leading models

SpaceXAI's Grok 4.6 achieves frontier-level performance (94.9% GPQA, 76.8 coding index) with 500K context and 80 tps throughput, at a fraction of the cost of Claude Opus 5 and GPT-5.6 Sol. API pricing is $2/$6 per 1M tokens with caching discounts. Strong agentic performance and turn efficiency on long-horizon tasks make it a compelling option for enterprise deployments, reinforcing the trend of cost-efficient frontier models. New benchmarks show Grok 4.6 matches GPT-5.6 Sol on Artificial Analysis at half the cost, though cost-per-task nuance means it's not always cheaper than Grok 4.5. Pricing remains same as Grok 4.5.