發布日期2026/07
推論速度0 tokens/s
首字延遲時間0.00s
輸入價格$2/1M
輸出價格$6/1M
排名表現
評測分析
Grok 4.5 is the strongest reasoning model currently available, well-suited for developers, scientists, and anyone requiring sharp analytical thinking. Its main limitation is weak autonomy, making it a poor choice for automation workflows or multi-step agent tasks.
強項
- Outstanding logical reasoning (#2 on AIM) — excels at solving complex problems and deep analysis
- Smooth general conversation (#5) — understands context and delivers flexible responses
- Competitive coding and debugging (#6) — solid support for developer workflows
弱項
- Limited AI agent (autonomous) capabilities (#22) — weak in complex automation and orchestration tasks
- Unclear response latency and context length limits compared to peers in its class
- Lack of competitive pricing information and SLAs compared to alternatives
適用情境
Solving logic puzzles, mathematical reasoning, and data analysis requiring rigorous thinkingWriting, reviewing, and debugging code; explaining algorithms and system architecturesIn-depth Q&A, brainstorming, and educating on complex technical concepts
指南與影片
Access Grok 4.5 via the x.ai website, Grok Build (AI code editor), or integrate it into Cursor IDE using an API key. The model is optimized for coding, agentic tasks, and knowledge work at a low cost (just $2/million input tokens, $6/million output tokens). For API access, obtain a key from console.x.ai and get started with a few lines of Python or JavaScript.