SpaceXAI released Grok 4.5 on July 8 at $2 per million input tokens and $6 per million output tokens, positioning it as a cost-competitive alternative in the AI model market. Elon Musk stated on X that the model is roughly comparable to Anthropic's previous flagship Opus 4.7 while emphasizing speed and cost advantages over benchmark performance. The release follows SpaceX's xAI merger closure in February and represents SpaceXAI's strategy to target developers, engineers, and knowledge workers with a western-developed model priced significantly below competitors like Anthropic's Claude Opus 4.8 and OpenAI's GPT models.
SpaceXAI Prices Grok 4.5 Below Competitor Models
Grok 4.5 costs $2 per million input tokens and $6 per million output tokens. Claude Opus 4.8, Anthropic's primary flagship, runs $5 input and $25 output. GPT 5.6 Sol, OpenAI's new top-tier model that also launched Wednesday, is priced at $5 input and $30 output. Musk posted on X that Grok 4.5 is roughly comparable to Opus 4.7 but much faster, framing the positioning as a deliberate tradeoff between speed and cost versus raw capability.
Grok 4.5 Benchmark Results Show Mixed Performance
SpaceXAI published four benchmark results at launch. On DeepSWE 1.1, which measures how reliably an AI can close real software bugs, Grok 4.5 scored 53%, behind Claude Opus 4.8 at 59% and GPT 5.5 at 67%. Claude Fable 5 topped the chart at 70%. On SWE Bench Pro, Grok 4.5 posted 64.7%, beating GPT 5.5's 58.6%. Opus 4.8 leads at 69.2%, and Fable 5 sits at 80.4%. The company's benchmarks compare against GPT 5.5, not GPT 5.6, because GPT 5.6 launched Wednesday hours after Grok 4.5's announcement.
SpaceXAI Trained Model on Nvidia GB300 GPU Infrastructure
SpaceXAI trained Grok 4.5 in collaboration with the recently acquired Cursor AI on tens of thousands of Nvidia GB300 GPUs inside Colossus, the Memphis supercomputer with total capacity across more than 200,000 GPUs. The model was trained on developer session data from Cursor, including debugging traces and real code edits rather than static repositories. The model runs at 80 tokens per second and is available via API on Hermes and Grok build with half a million tokens of context.
Grok 4.5 Uses Fewer Tokens Per Task Than Opus 4.8
On SWE Bench Pro tasks, Grok 4.5 used an average of 15,954 output tokens to complete each job. Opus 4.8 burned through 67,020 tokens for the same work, a 4.2x gap. The model is not available in the EU yet; SpaceXAI says European access is expected in mid-July.
FAQ
What did SpaceXAI release on July 8?
SpaceXAI released Grok 4.5 on July 8, priced at $2 per million input tokens and $6 per million output tokens. Elon Musk stated the model is roughly comparable to Anthropic's Opus 4.7 while emphasizing speed and cost advantages.
How does Grok 4.5 perform on software engineering benchmarks?
Grok 4.5 scored 53% on DeepSWE 1.1 and 64.7% on SWE Bench Pro. On DeepSWE 1.1, it trails Claude Opus 4.8 at 59% and GPT 5.5 at 67%. On SWE Bench Pro, it beats GPT 5.5's 58.6% but remains below Opus 4.8's 69.2%.
When will Grok 4.5 be available in the EU?
Grok 4.5 is not available in the EU yet. SpaceXAI says European access is expected in mid-July.
