SpaceXAI has released Grok 4.5, its first major AI model since becoming a public entity under Elon Musk's SpaceX, positioning it as a fast, low-cost option for coding and agent-based tasks. The model was trained in collaboration with Cursor, an AI-powered coding platform SpaceXAI is reportedly acquiring for $60 billion, using real-time debugging traces and live developer sessions instead of static code datasets. Priced at $2 per million input tokens and $6 per million output tokens, Grok 4.5 undercuts competing models significantly. SpaceXAI claims it matches Anthropic's Opus 4.7 in performance while running faster, though Opus 4.7 is not the current flagship—having been surpassed by Opus 4.8 and the newer Claude Fable 5.
Benchmark results show Grok 4.5 trailing top performers in raw capability. On DeepSWE 1.1, it resolved 53% of real-world bugs, below Opus 4.8's 59%, GPT-5.5's 67%, and Claude Fable 5's 70%. On SWE-Bench Pro, Grok scored 64.7%, ahead of GPT-5.5's 58.6% but behind Opus 4.8 at 69.2% and Fable 5's 80.4%. Notably, SpaceXAI compared its model to GPT-5.5, not the newly launched GPT-5.6, which arrived just hours later. Where Grok excels is efficiency: it used only about 15,954 output tokens per SWE-Bench Pro task, less than a quarter of the 67,020 used by Opus 4.8. Combined with its speed of approximately 80 tokens per second, this efficiency could make Grok 4.5 cheaper to operate at scale despite lower output quality.
OpenAI rolled out GPT-5.6 the following day, offering three tiers—Sol, Terra, and Luna—with input/output pricing at $5/$30, $2.50/$15, and $1/$6 per million tokens respectively. Sol, the flagship tier, showed improved chain-of-thought controllability over GPT-5.5. The release followed a limited preview restricted to government users, which OpenAI attributed to safety testing by the U.S. Commerce Department's Centre for AI Standards and Innovation. CEO Sam Altman stated that such government-gated rollouts are not a sustainable model for future launches. Both companies framed their models as challengers to Anthropic's Opus line, even as neither surpassed the current leaders in benchmark performance.
Elon Musk claims Grok 4.5 competes with Opus-class models, but the comparison hinges on a version already outdated at launch. The model's real advantage lies in cost and speed, not capability, suggesting a strategic pivot toward volume-driven use cases over cutting-edge performance. For Nigerian developers relying on affordable AI tools, lower token prices could matter more than benchmark rankings. However, the $60 billion Cursor acquisition deal remains unconfirmed by either company, casting uncertainty over the collaboration's scale.
Editorial note: AI-assisted opinion, not established fact. Full disclaimer →
