Grok 4.5 promises high performance at a fraction of token costs

SpaceXAI has released Grok 4.5, its first major model update since going public, positioning it as a high-performance “Opus-class” system according to Elon Musk. The company claims Grok 4.5 delivers twice the token efficiency of competing models, potentially lowering costs for coding, research, writing, and other knowledge-work tasks. Benchmark results show Grok 4.5 performing competitively with top-tier models, though slightly below the best in class, with Musk stating it is roughly comparable to Anthropic’s Opus 4.7 but faster and cheaper. SpaceXAI priced the model at $2 per million input tokens and $6 per million output tokens, significantly undercutting Anthropic and OpenAI’s premium offerings. The release arrives during a busy week for AI, with OpenAI preparing to launch GPT 5.6, described as its strongest model yet.

Read the full story on TechCrunch →