llmrelay
// other

Grok 4.5

xAI reasoning model with a 500K context window.

Grok 4.5 is xAI’s reasoning model, positioned as an alternative to Claude and GPT for long-document analysis and codebase-scale reasoning. It carries a 500k context window - half the size of Claude’s or GPT’s, but still large enough for most multi-file work - and costs less per token than GPT-5.6 Terra or Claude Sonnet 5.

xAI markets Grok as tuned for reasoning transparency, with the option to request step-by-step traces. That makes it a useful second opinion on work done by Claude or GPT, and a way to verify that a model actually thought through the problem rather than pattern-matched to an answer.

We sell it at half xAI’s list price - $1.00 per million input tokens against $2.00. Note that xAI’s official pricing has two tiers: under 200k tokens it is $2/M, at or above 200k it jumps to $4/M. We charge the lower rate regardless of prompt length, which makes long-context work on Grok cheaper here than directly through xAI.

Previous generation

Grok 4.5 still runs on llmrelay at the rates below, and we keep selling it — pinned versions matter when you have output you do not want to re-validate. For new work, Grok 4.6 is the current pick at $1/M input.

Official list
llmrelay
Input / M tokens
$2.00
$1.00
Output / M tokens
$6.00
$3.00
Context window
—
500,000 tokens
Max output
—
32,768 tokens

What it costs you per month

Real-world budget scenarios. Numbers are simple sums — official list price vs llmrelay's 50% off tier.

Usage scenario
Official
llmrelay
Light coding (1M in / 200K out per month)
$3.20
$1.60save $1.60
Heavy Cursor / Cline user (50M in / 5M out per month)
$130.00
$65.00save $65.00
Production RAG (500M in / 20M out per month)
$1120.00
$560.00save $560.00

What it's good at

Best for

Pick Grok 4.5 when

  • +You want a second opinion outside the Claude/GPT duopoly. Running the same task through Grok catches model family blind spots.
  • +Long-document reasoning on a budget. At $1.00/M input Grok is cheaper than Claude Sonnet 5 and half the cost of GPT-5.6 Terra, while still carrying a 500k window.
  • +You value reasoning transparency. Grok’s step-by-step trace mode is useful for understanding how it reached a conclusion, which the other families do not expose as cleanly.
  • +Whole-repo code review where 500k context is enough. That window fits most repositories, and the per-token savings over Claude or GPT add up fast on large codebases.

Choose something else when

  • !You need more than 500k context. Claude Opus, Claude Sonnet, GPT-5.6 Sol and GPT-5.6 Terra all carry 1M windows, which is twice what Grok offers.
  • !Tool use and function calling are load-bearing. Grok supports it, but Claude’s and OpenAI’s implementations are more mature and more widely integrated into third-party tools.
  • !You are locked into a specific ecosystem. If your workflow is built around Claude Code or OpenAI function calling, switching to Grok for cost savings alone is not worth the friction.
  • !Ordinary copy. Grok is a reasoning model. For hard Claude work use Opus 5 or Fable 5.1, not a writing SKU.

Questions people ask about Grok 4.5

How much does Grok 4.5 cost?

xAI’s list price has two tiers: under 200k input tokens it is $2.00 per million input and $6.00 per million output; at or above 200k input it is $4.00 and $12.00. On llmrelay we charge the lower tier regardless of prompt length - $1.00 and $3.00 - which makes long-context work on Grok cheaper here than buying directly from xAI.

Should I use Grok 4.5 or Claude Sonnet 5?

Grok costs $1.00/M input against Sonnet’s $1.00, so they are equivalent on price. Sonnet has a 1M context window; Grok has 500k. Sonnet has wider tool integration; Grok has reasoning traces. If 500k context is enough and you value transparency, Grok is a strong pick. If you need the full 1M window or are already in the Claude ecosystem, Sonnet is safer.

What is the context window on Grok 4.5?

500,000 tokens input, with up to 32,768 output tokens. That is half the window of Claude Opus, Claude Sonnet, or GPT-5.6 models (all 1M), but still large enough for most multi-file code reviews or long-document analysis.

Can I use Grok 4.5 with Cursor or Claude Code?

Yes, if the tool lets you set a custom base URL. Point it at https://api.llmrelay.dev/v1 and use the model id grok-4.5. That said, Cursor and Claude Code are optimised for their native model families, so you may hit rough edges using Grok instead.

Try Grok 4.5 at half the price

Free to create an account, no subscription, $10 minimum top-up — enough to run this model against a real task and judge quality yourself.

Get API key →

Compare Grok 4.5 against the alternatives

The comparisons this model appears in, the models nearest it on price, and the full rate card.