SpaceXAI Releases Grok 4.5, a Cursor-Trained Model for Coding, Agentic Tasks, and Knowledge Work at $2/M Input

SpaceX's AI division has unveiled Grok 4.5, a large language model specifically optimized for coding, agentic tasks, and professional knowledge work. The release marks a significant competitive move in the increasingly crowded market of specialized AI models, positioning Grok 4.5 as an accessible alternative to existing solutions through aggressive pricing and specialized training methodologies.

Grok 4.5 has been specifically trained using data from Cursor, a popular AI-powered code editor, which enhances its ability to handle programming tasks and development workflows. The model operates at 80 tokens per second throughput, providing reasonable speed for real-time applications. Pricing is notably competitive at $2 per million input tokens and $6 per million output tokens, undercutting several established competitors in the market.

The model has achieved top-tier performance benchmarks, notably ranking first on Harvey's Legal Agent Benchmark—a specialized evaluation designed to assess AI performance in legal domain tasks. This achievement suggests the model's strength extends beyond coding into complex knowledge-based work requiring domain expertise and reasoning capabilities.

• Pricing Pressure: The $2 input token pricing creates downward pressure on the broader market, potentially forcing competitors to reconsider their pricing strategies for similar capabilities

• Cursor Integration: Deep optimization for Cursor-based workflows addresses a specific developer community, potentially driving adoption among this significant user segment

• Diversified Applications: Success on legal benchmarks indicates broader applicability across professional sectors beyond software development, expanding addressable markets

• Competitive Consolidation: The release reinforces SpaceX's commitment to AI infrastructure, competing directly with OpenAI, Anthropic, and other specialized model providers

• Agentic AI Focus: Emphasis on agentic task capabilities positions Grok 4.5 for the emerging autonomous AI agent market

The launch of Grok 4.5 demonstrates how established technology companies are entering AI markets with aggressive strategies targeting specific use cases and price points. For developers, enterprises evaluating AI tools, and knowledge workers, this release provides another high-performance option at competitive pricing. The model's specialized training approach and impressive benchmark performance suggest that future AI model differentiation will increasingly depend on domain optimization rather than general capability alone. This trend will likely accelerate innovation while fragmenting the market into specialized solutions tailored to specific professional contexts.