grok-4.5 API — per-token pricing and code examples
Price per 1M tokens
Cache rates apply when the provider reports cache tokens.
90% offAbout this model
grok-4.5 is served through OpenAI Chat Completions, OpenAI Responses, Anthropic Messages. You pay $0.20 per 1M input tokens and $0.60 per 1M output tokens. That is 90% below the official price of $2.00 and $6.00. Cache reads cost $0.02 per 1M tokens when the provider reports cached tokens. It is included in budget packs from $9.99 up.
Where Grok 4.5 fits
Grok 4.5 is the coding model from xAI, described by its maker as built "for agentic software, engineering, and workflow tasks." xAI lists a context window of 500,000 tokens for grok-4.5, enough to hold a sizeable slice of a repository plus the conversation around it. The model reads text and images and writes text, so a screenshot of a broken layout or a failing CI page can travel in the same request as the log. xAI does not publish a fixed maximum output length on the Grok 4.5 model page, so if very long answers would hurt your app, cap the output yourself with the token limit field of the endpoint you call. Function calling and structured outputs are both marked as supported, which makes grok-4.5 a reasonable choice for agents that drive tools and for pipelines that expect JSON in a known shape.
Grok 4.5 cannot skip its reasoning step
According to xAI, reasoning on Grok 4.5 cannot be disabled. Every request spends tokens thinking before the visible reply begins, even a one-line question. What you do control is the effort level: for grok-4.5, xAI documents low, medium and high, and when the setting is missing xAI applies high. A lower level is a sensible start for routine edits, renames and quick triage, while high earns its keep when Grok 4.5 has to follow a bug through several files.
xAI groups Grok 4.5 with its reasoning models and says those models reject presence penalty, frequency penalty and stop sequences: a request that includes any of them returns an error. Client code written years ago for plain chat models often sets one of these fields by default, so strip them before you point that code at grok-4.5.
What it costs on Cent AI
Grok 4.5 is billed per token from your prepaid USD balance, using the input, cache read, cache write and output rates in the price box on this page. Since Grok 4.5 always reasons, each call carries reasoning tokens, and Cent AI bills those as output just like the text you see. On this model the effort level is the first knob to turn when spend climbs: moving simple jobs from high to low can shorten the hidden thinking as well as the wait.
xAI lists a cached input price for grok-4.5, so a repeated prompt prefix can be served from cache. Cent AI applies cache rates only when xAI's usage report for that request contains cached tokens, and never estimates hits by itself. xAI's caching guide recommends sending a stable prompt_cache_key on the Responses API so related calls reach the same cache, which pays off most in agent loops that resend a long system prompt and tool list every turn.
Using it from your tools
You can call grok-4.5 on all three Cent AI endpoints: /v1/chat/completions, /v1/responses and /v1/messages. Cent AI does not change your parameters silently, so a field that xAI refuses for Grok 4.5 is not dropped and comes back as an upstream_error with the model's status code. xAI documents the effort setting for the Responses API as a reasoning object; the example below uses the /v1/responses format.
{
"model": "grok-4.5",
"reasoning": { "effort": "low" },
"prompt_cache_key": "repo-triage-run-42",
"input": "List the files touched by this stack trace."
}
Keep the prompt_cache_key value fixed for one conversation or one agent run, and change it when the task changes. If you move a working setup from another OpenAI-compatible model to grok-4.5, check its defaults for penalty and stop fields first.
Specification
- Context window
- —
- Input and output
- Text
- Endpoints
- /v1/chat/completions, /v1/responses, /v1/messages
- Smallest pack
- $9.99
Latency and error rate per model are on the status page. status.cent-ai.dev
What a request costs
Open the full calculatorCost per request
Estimate
grok-4.5An estimate at price version 2. Your bill comes from the usage the provider reports for each request.
Save $0.4816Code
curl https://api.cent-ai.dev/v1/chat/completions \
-H "Authorization: Bearer $CENT_AI_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "grok-4.5", "messages": [{"role": "user", "content": "Hello"}]}'Set CENT_AI_KEY to a key from the dashboard. OpenAI SDKs use https://api.cent-ai.dev/v1; Anthropic SDKs use https://api.cent-ai.dev.
Price history
Version 2, in effect since Nov 6, 2026.
Effective
per 1M tokens
Inputold $0.10 new $0.20Cache readold $0.01 new $0.02Cache writeold $0.10 new $0.20Outputold $0.30 new $0.60
Compare with
- hy4-previewInput$0.46704
- deepseek-v4-proInput$0.165
- gpt-6.1-solInput$0.58
- gpt-6.1-terraInput$0.15
Questions
How much does grok-4.5 cost on Cent AI?
$0.20 per 1M input tokens and $0.60 per 1M output tokens. Cache reads cost $0.02 and cache writes $0.20 per 1M tokens. Each request is rounded up once.
Is there a minimum charge?
A successful request costs at least $0.00002. A request that ends early, because the client left or the output was cut off, pays only for its tokens.
How do I pay?
Top up a prepaid USD balance with QRIS or USDT/USDC, or buy a budget pack once. Nothing renews automatically.
What if the provider does not report usage?
The request is billed from an estimate and marked estimated on the Requests page. An estimate never contains cache tokens.
Which endpoint do I call?
/v1/chat/completions, /v1/responses, /v1/messages. The OpenAI base URL is https://api.cent-ai.dev/v1.
Try grok-4.5 with one key
One prepaid balance for every model. Pay with QRIS or USDT/USDC.