GPT-5.3 Codex

GPT-5.3 Codex by OpenAI: $1.75 input and $14 output per 1M tokens, 400K context. Call it through the odnoga LLM gateway.

ReasoningVisionToolsJSON modeStreaming

GPT-5.3 Codex is served by OpenAI and called through odnoga with the same OpenAI-compatible request shape as every other model in the catalog. It is a reasoning model, so it spends extra output tokens working through a problem before answering — budget for higher output cost on hard tasks. It accepts images alongside text. It supports tool and function calling. It can be forced to return structured JSON. Its 400K-token context window sets how much input you can send in one request. Cached input is billed at $0.175 per 1M tokens, so repeated prefixes cost less.

Specification and price

VendorOpenAI
Model IDgpt-5.3-codex
Context window400K tokens
Max output128K tokens
CapabilitiesReasoning, Vision, Tools, JSON mode, Streaming
Your plan
Per 1M tokensVendor costYour price on Free+7%
Input$1.75$1.88
Output$14$15.05
Cached input$0.175$0.188

Vendor cost is the list price per million tokens as recorded in the odnoga catalog; your price applies your plan margin with the same formula that bills every request — see pricing. Pricing

Call it through odnoga

const res = await fetch("https://api.odnoga.com/v1/chat/completions", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.ODNOGA_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    model: "gpt-5.3-codex",
    messages: [{ role: "user", content: "Hello" }],
  }),
});

Same request shape for every vendor in the catalog — swap the model id and odnoga handles keys, routing, limits and cost accounting.

How to use GPT-5.3 Codex

Reviewed 2026-09-14

Best for

  • Agentic coding: multi-file changes, refactors, test writing and long-running tasks in a Codex-style environment.
  • Work that mixes research, tool use and execution over many steps rather than a single answer.
  • Code review and bug hunting where the model needs the whole repository in context (400,000-token window).
  • Pipelines where you steer the model while it works instead of waiting for one final output.

Not the right pick when

  • General chat, writing or analysis — this model is tuned for agentic coding, not as a general assistant.
  • Short, cheap code completions: a smaller general model is usually enough.
  • Environments without sandboxed tool execution, where most of its advantage cannot be used.

Practical tips through odnoga

  1. 01Follow the Codex prompting guide: describe the goal, the constraints and how done is verified, then let the model run the loop.
  2. 02Use reasoning effort as the budget dial — the model supports low through xhigh.
  3. 03Long agent runs are where spend escapes: set a per-tenant budget and an alert before you let it run unattended.
  4. 04Record a real run with odnoga capture and freeze it as a test case, so the next model upgrade is compared on the work you actually do.

What a month costs

1,000 calls a month, 10,000 input tokens and 2,000 output tokens each, at your Free price (vendor cost +7%):

Input (10M tokens)$18.82
Output (2M tokens)$30.11
Your cost per month on Free$48.92

Vendor list cost $45.50 + odnoga margin $3.42 (+7%). Cached input or answer reuse lowers it; odnoga records both numbers per request.

GPT-5.3 Codex compared

ModelContextInput / 1MOutput / 1MCapabilities
GPT-5.3 Codex400K$1.75$14Reasoning, Vision, Tools, JSON mode, Streaming
ChatGPT (chat-latest)400K$5$30Vision, Tools, JSON mode, Streaming
GPT Image 1 mini$2$8Vision

Questions

What is GPT-5.3-Codex optimized for?
OpenAI describes it as optimized for agentic coding tasks in Codex or similar environments, with low, medium, high and xhigh reasoning effort settings and a 400,000-token context window.
Should I use it for everything in a developer tool?
No. Use it for the agent loop and a cheaper general model for chat, summaries and metadata around it. odnoga lets both live behind the same endpoint with separate prompts and budgets.
How do I keep an agent run from overspending?
Cap reasoning effort, set a per-tenant budget in odnoga, and watch the per-request ledger — every step is recorded with its own cost.

All models · Pricing · Docs · Compare models in the evaluation lab

Sources: OpenAI — GPT-5.3-Codex model page, OpenAI — Codex prompting guide

One gateway, every model.