SpaceXAI releases Grok 4.5; Musk claims Opus-class performance at half the cost

SpaceXAI released Grok 4.5 on July 8, its first major model launch post-IPO and post-Cursor acquisition, with Elon Musk claiming Opus-class performance at approximately half the inference cost of competing models. Grok 4.5 is priced at $2 per million input tokens and $6 per million output tokens—compared to Claude Opus 4.8 at $5 input / $25 output and GPT 5.6 at $5 input / $30 output. Musk's internal assessment rates Grok 4.5 as "roughly comparable to Opus 4.7, but much faster," with 4.2x greater token efficiency on SWE Bench Pro tasks (15,954 output tokens average per task vs. Opus 4.8's 67,020).

On benchmark results, Grok 4.5 ranks fourth on Artificial Analysis Intelligence Index with a score of 54 (behind Fable 5, GPT-5.5, and Opus 4.8) and achieves on-par coding agent performance with GPT-5.5 at significantly lower cost. The model is trained on Cursor development session data and runs on tens of thousands of NVIDIA GB300 GPUs. Available today in Grok Build, Cursor (all plans), and the SpaceXAI console; EU availability deferred to mid-July.

For teams evaluating coding agents and knowledge work models, Grok 4.5's appeal is pure economics: fourth-tier capability with frontier token efficiency and sub-$1 per-task cost puts it squarely on the cost-performance Pareto frontier. That positioning—cheaper-than-Opus with higher throughput—replicates the playbook DeepSeek deployed successfully in the Chinese market. Watch for adoption signals among engineering teams already locked into Cursor, where the model ships by default.