GPT-5.3-Codex · OpenAI

What does GPT-5.3-Codex cost to run in a coding agent?

GPT-5.3-Codex costs about $0.055 per average coding-agent request on our measured traffic, at list prices of $1.75 per million input and $14 per million output tokens. It scores 33 on the Artificial Analysis Intelligence Index (an estimate), 27th of the 48 models we track. Source: Artificial Analysis (artificialanalysis.ai)

List prices: OpenAI API 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

Run current Claude, GPT, Gemini and open models 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 GPT-5.3-Codex costs per coding-agent request

Cost at list price	GPT-5.3-Codex
Prompt under 100K tokens50% of requests, 87% of input cached	$0.030
Prompt of 100K to 272K39% of requests, 88% of input cached	$0.073
Prompt over 272K tokens11% of requests, 86% of input cached	$0.13
Average request	$0.055
100 requests	$5.55
Median session16 requests; a session ends after 10 idle minutes	$0.74
A month of regular usemedian of 20 users active on 5+ days	$84.36

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. 3% of requests were larger than GPT-5.3-Codex's 400K context window and are left out of its figures.

GPT-5.3-Codex benchmarks, speed and latency

Artificial Analysis	Value	Rank
Intelligence Index10 evaluations, 0 to 100	33 (estimate)	27 of 48
Output speedmedian	113 tokens/s	16 of 44
Time to first answer tokenincludes thinking time	72 s	26 of 44

Source: Artificial Analysis (artificialanalysis.ai). Measured variant: GPT-5.3 Codex (Xhigh), speed on OpenAI. 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.

About GPT-5.3-Codex

GPT-5.3-Codex is a GPT-5.3 variant OpenAI tuned for software engineering, released February 5, 2026 for sandboxed execution, diff generation and long coding sessions. Codex CLI used to default to it; OpenAI has since deprecated it in Codex with ChatGPT sign-in. It has a 400K context and lists at $1.75/$14 per million tokens. Artificial Analysis estimates 33 on its Intelligence Index at about 113 output tokens per second and has no Terminal-Bench 4.0 result for it. GPT-6.1 Sol, OpenAI's current Codex pick, scores 52.

Sources: Artificial Analysis

GPT-5.3-Codex FAQ

How much does GPT-5.3-Codex cost per coding-agent request?

$0.055 on average at list price, $5.55 per 100 requests, and $0.74 for a median 16-request session. That is 79,301 real coding-agent requests (Sep 13 to Oct 12, 2026) priced at OpenAI'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 GPT-5.3-Codex charge more for long prompts?

No. GPT-5.3-Codex bills the same rate per token up to its 400K-token context window, which is smaller than 3% of the coding-agent requests we measured.

How fast is GPT-5.3-Codex?

Artificial Analysis measures GPT-5.3 Codex (Xhigh) at 113 output tokens per second, with 72 seconds to the first answer token, thinking time included, on OpenAI's API. Source: Artificial Analysis (artificialanalysis.ai).

Is GPT-5.3-Codex good for AI agents?

Existing Codex setups on an API key that still pin GPT-5.3-Codex. On our scoring, which is calibrated to public benchmarks, it rates 5.1/10 for agentic ability and 6.8/10 for output quality. Tuned by OpenAI for coding and tool use.

What is GPT-5.3-Codex's context window?

GPT-5.3-Codex has a 400K-token context window. OpenAI is the maker.

Can I self-host GPT-5.3-Codex?

No. GPT-5.3-Codex is a proprietary, hosted-only model from OpenAI. You access it through an API, such as OpenRouter, rather than running the weights yourself.

What are the downsides of GPT-5.3-Codex?

The main trade-offs: Deprecated in Codex with ChatGPT sign-in; GPT-6.1 Sol scores far higher for about 20% more per request ($0.067 vs $0.055). It's strongest for existing Codex setups on an API key that still pin GPT-5.3-Codex.