xAI Lists Grok 4.5: $2 Input, $6 Output, 500k Context Window

What's New

xAI quietly added Grok 4.5 to the model roster on 2026-07-08. It accepts text, images, and files as input and returns text — a practical multimodal setup for document-heavy or mixed-media pipelines.

Pricing at a Glance

• Input: $2.00 / 1M tokens
• Output: $6.00 / 1M tokens

The 3:1 output-to-input ratio is fairly standard for this tier. For a workload that's 80% input / 20% output — say, long-document summarisation — your blended cost comes in around $2.80 per million tokens. Heavier generation workloads (code, reports) will push that closer to the $6.00 ceiling.

The 500k Context Window

A 500k-token context is the headline capability here. That's enough to load in several large PDFs, a substantial codebase, or an extended conversation history without chunking. Whether that matters to you depends entirely on your retrieval strategy — if you're already running RAG effectively, the extra headroom may not move the needle.

Who Should Pay Attention

• Developers working with long documents or large file inputs get a native multimodal path without preprocessing gymnastics.
• Cost-sensitive teams should benchmark against comparable models in our index before committing — $6.00 output is competitive but not the cheapest option available.
• Existing xAI users on Grok 3.5 or earlier will want to weigh whether the larger context justifies the switch.

Check live specs, context limits, and price history on the Grok 4.5 — live specs & price history page before finalising your stack.

Ezra, Scout AI Team