grok-4.5

By price, grok-4.5 ranks 48th cheapest of 68 on input and 45th on output. Its quality score of 91 ranks 17th of 68, and on value (quality score ÷ output price) it ranks 43rd of 68. Its blended rate sits 2% above the catalog average.

In real billing terms: a support chatbot handling 50,000 turns a month (2,000 input / 500 output tokens each) costs about $350 per month on grok-4.5, while a RAG workload of 20,000 queries (8,000 input / 800 output tokens each) runs about $416. Because output is priced 3.0× input here, long-generation tasks scale cost fastest — cap output length before you optimise anything else.

A 500K context window holds roughly 667 pages of A4 text (at ~750 tokens per page), ranking 32nd of 68 here. That is enough to drop in an entire technical manual or several source files at once, which suits long-document summarization and codebase-wide analysis. Output runs at about 91 tok/s (39th fastest of 68), acceptable interactively, though long responses will feel slow.

We track 3 xAI models, and grok-4.5's blended rate sits 47% above that provider's average. If your task tolerates a little less headroom, qwen3.5 (Alibaba) blends about 96% cheaper at a quality score of 91, and is worth benchmarking first. Going the other way, deepseek-reasoner-v4 scores 92 on quality for about 82% less.

The closest model in xAI's line-up is grok-4.20 ($1.25 input / $2.50 output, quality —). Against it, grok-4.5 scores 91 points higher on quality, has a 1500K smaller context window, costs 48% more blended. When two models sit this close on price, the deciding factor is usually the quality ceiling — run both against your real prompts before picking.