Grok 4.3 · xAI

What does Grok 4.3 cost to run in a coding agent?

Grok 4.3 costs about $0.068 per average coding-agent request on our measured traffic, at list prices of $1.25 per million input and $2.50 per million output tokens. It scores 25 on the Artificial Analysis Intelligence Index, 36th of the 48 models we track. Source: Artificial Analysis (artificialanalysis.ai)

List prices: xAI models and pricing, checked 2026-10-13. Request costs: Standard Compute gateway data, 79,301 coding-agent requests from 75 users, Sep 13 to Oct 12, 2026. How we compute this

A newer version is out. Grok 4.7 →

Run Grok 4.3 in your agent

Or let the router send each step to the cheapest model that can do it, on providers we vet for price, uptime and speed. Plans from $19 a month.

See plans

What Grok 4.3 costs per coding-agent request

Cost at list price	Grok 4.3
Prompt under 100K tokens50% of requests, 87% of input cached	$0.021
Prompt of 100K to 272K39% of requests, 88% of input cached	$0.077
Prompt over 272K tokens11% of requests, 86% of input cached	$0.26
Average request	$0.068
100 requests	$6.84
Median session16 requests; a session ends after 10 idle minutes	$0.61
A month of regular usemedian of 20 users active on 5+ days	$103

Each logged request priced at the vendor’s list rates for its own token counts, including cache reads and any higher rate for long prompts. Average output: 698 tokens per request.

Grok 4.3 benchmarks, speed and latency

Artificial Analysis	Value	Rank
Intelligence Index10 evaluations, 0 to 100	25	36 of 48
Terminal-Bench 4.0agentic coding in a terminal	0.0%	35 of 39
Output speedmedian	113 tokens/s	17 of 44
Time to first answer tokenincludes thinking time	25 s	14 of 44

Source: Artificial Analysis (artificialanalysis.ai). Measured variant: Grok 4.3 (High), speed on SpaceXAI. Fetched October 13, 2026. Rank is among the models on this site, older versions included. Artificial Analysis runs Terminal-Bench 4.0 in its own setup, so its scores differ from the tbench.ai leaderboard.

What to watch for

Grok 4.7 has replaced it and costs $0.13 per request on our traffic. Prompts 200K tokens or more (21% of our requests) bill at $2.50 input and $5 output per million. It is weak on long agentic coding tasks (see Terminal-Bench 4.0 above), so a cheap request can turn into more steps and retries.

About Grok 4.3

Grok 4.3 was xAI's flagship from April 30, 2026, with a 1M-token context at $1.25/$2.50 per million tokens. It scores 25 on the Artificial Analysis Intelligence Index and 0% in Artificial Analysis's Terminal-Bench 4.0 run, so it is not a fit for long agentic coding. Grok 4.7 replaced it in September 2026 and scores 46. Prompts of 200K tokens or more bill at higher rates.

Sources: Artificial Analysis

Grok 4.3 FAQ

How much does Grok 4.3 cost per coding-agent request?

$0.068 on average at list price, $6.84 per 100 requests, and $0.61 for a median 16-request session. That is 79,301 real coding-agent requests (Sep 13 to Oct 12, 2026) priced at xAI's list rates; the median request sent 99,534 input tokens, 87% of input was read from cache, and requests returned 698 output tokens on average.

Does Grok 4.3 charge more for long prompts?

Yes. Prompts 200K tokens or more are billed at $2.50 per million input and $5 per million output tokens, on every token in the request including cache reads. 21% of the coding-agent requests we measured were that large.

How fast is Grok 4.3?

Artificial Analysis measures Grok 4.3 (High) at 113 output tokens per second, with 25 seconds to the first answer token, thinking time included, on SpaceXAI's API. Source: Artificial Analysis (artificialanalysis.ai).

Is Grok 4.3 good for AI agents?

Simple, long-context calls in existing Grok setups. On our scoring, which is calibrated to public benchmarks, it rates 4.4/10 for agentic ability and 5.7/10 for output quality. 1M-token context.

What is Grok 4.3's context window?

Grok 4.3 has a 1M-token context window, large enough to hold sizeable codebases or document sets in a single run. xAI is the maker.

Can I self-host Grok 4.3?

No. Grok 4.3 is a proprietary, hosted-only model from xAI. You access it through an API, such as OpenRouter, rather than running the weights yourself.

What are the downsides of Grok 4.3?

The main trade-offs: Superseded by Grok 4.7, which scores far higher; 0% in Artificial Analysis's Terminal-Bench 4.0 run; Prompts of 200K tokens or more bill at higher rates. It's strongest for simple, long-context calls in existing Grok setups.