SpaceXAI Launches Grok 4.5, Challenges Rivals with Aggressive Pricing

SpaceXAI has officially launched Grok 4.5, a new large language model that Elon Musk touts as an "Opus-class model" offering superior speed, token efficiency, and a significantly lower cost compared to its rivals. The model, released on Wednesday, July 8, 2026, and becoming publicly available on Thursday, directly challenges leading AI models from Anthropic and OpenAI in the fiercely competitive enterprise AI market.

Built on a 1.5-trillion-parameter V9 foundation model and enhanced with supplemental training from the recently acquired Cursor, Grok 4.5 is specifically engineered for demanding tasks in coding, agentic operations, and knowledge work across finance and legal sectors. Its aggressive pricing strategy positions it as a major disruptor, costing $2 per million input tokens and $6 per million output tokens. This dramatically undercuts competitors like Anthropic's Opus 4.8, which is priced at $5 per million input and $25 per million output, and OpenAI's GPT-5.5/5.6 at $5 and $30 respectively. Additionally, SpaceXAI claims Grok 4.5 achieves roughly double the token efficiency of comparable models, solving tasks in fewer steps and delivering results at 80 tokens per second.

While internal benchmarks show promising performance, such as nearly matching GPT 5.5 on Terminal Bench 2.1, some independent evaluations indicate Grok 4.5 lags behind on more complex software engineering tasks like DeepSWE 1.1 and SWE Bench Pro. Nevertheless, its specialized training, particularly from Cursor data, and its focus on autonomous, multi-step tasks in professional domains, signal a strategic play by SpaceXAI to capture a significant segment of the enterprise market. The model's competitive cost structure and efficiency gains could compel businesses to re-evaluate their AI infrastructure, potentially driving down overall AI operational expenses and accelerating the adoption of advanced AI for complex workflows, despite the mixed benchmark results.