SpaceXAI launches Grok 4.5 to take on Claude and GPT-5.6

TL;DR:

  • SpaceXAI has launched Grok 4.5, its most capable model yet, aimed at coding and agentic work.
  • Elon Musk called it “an Opus-class model, but faster, more token-efficient and lower cost” than Anthropic’s Claude Opus 4.8.
  • It arrives days before OpenAI’s delayed public launch of GPT-5.6, intensifying the frontier price war.

SpaceXAI has launched Grok 4.5, calling it the company’s most intelligent model to date and pitching it squarely at software developers. Trained across tens of thousands of Nvidia GB300 GPUs, the model is built for coding and agentic tasks and is available immediately through SpaceXAI’s Grok Build coding agent, inside Cursor, and via the company’s developer console.

An Opus-class model at a fraction of the price

The launch is as much about economics as capability. Grok 4.5 is priced at $2 per million input tokens and $6 per million output tokens. That undercuts Anthropic’s Claude Opus 4.8, at $5 and $25 respectively, while sitting close to OpenAI’s GPT-5.6 Luna at $1 and $6. “It is an Opus-class model, but faster, more token-efficient and lower cost,” Musk wrote on X. The distribution play is notable too: SpaceX agreed last month to buy Anysphere, the startup behind Cursor, in an all-stock deal worth $60bn (£47bn), giving SpaceXAI a direct route into the enterprise coding market rather than relying on rivals’ tools. Musk’s xAI was absorbed into SpaceX in February and rebranded SpaceXAI.

Looking forward

For UK developers and the businesses buying their output, the takeaway is a widening, cheaper field of frontier coding models rather than any single winner. Grok 4.5’s pricing keeps downward pressure on a market where Anthropic still commands a premium, and EU availability is expected mid-July. The timing sharpens the contest: OpenAI is due to launch its most advanced model, GPT-5.6, on Thursday after a government-forced delay over national-security concerns. With three “Opus-class” agents landing in a single week, the differentiator for enterprise buyers is shifting from raw capability towards cost, latency and how cleanly each model plugs into existing developer workflows.

Share this article