Grok 4.5 by SpaceXAI: Advanced AI for Coding, Engineering & Office Work
Discover Grok 4.5, SpaceXAI's smartest model yet. Excelling in coding, agentic tasks, and Microsoft Office integration, it offers fast 80 TPS speeds and unmatched token efficiency.
Key Takeaways
- Built for Complex Tasks: Grok 4.5 is SpaceXAI's smartest model designed specifically for coding, agentic tasks, and knowledge work, having been trained alongside Cursor.
- Engineering and Legal Excellence: It excels in real-world engineering, scoring highly across major benchmarks like DeepSWE, Terminal Bench 2.1, and SWE Bench Pro, while also taking the #1 spot on Harvey's Legal Agent Benchmark.
- High Speed and Efficiency: The model operates at fast-model speeds of 80 tokens per second (TPS). It is exceptionally efficient, requiring about 4.2 times fewer output tokens than Opus 4.8 (max) to resolve tasks on SWE Bench Pro.
- Office Work Proficiency: Grok 4.5 is the default model in Grok Build and is highly capable of handling Microsoft Office tasks, including building complex Excel models, designing PowerPoint presentations using native shapes, and writing clear prose in Word.
- Competitive Pricing: The model costs $2 per million input tokens and $6 per million output tokens, delivering high intelligence per unit of time and cost due to its token efficiency.
- Availability: Grok 4.5 is available today via Grok Build, Cursor, and the SpaceXAI API console, with European Union availability expected in mid-July 2026.
References
