← Model list

GPT-5.3 Codex

openai·openai/gpt-5.3-codex·GA·Closed·gpt-codex series

Coding-optimized GPT model for repository edits, reviews, and agentic software work

At a glance
Official price$1.75 / $14.00 per 1M
Cache read $0.175 · OpenAI
Lowest paid$1.57 / $12.60
Requesty Gateway · 1.4× spread
Context400,000
Output limit128,000
Capabilities
ReasoningTool useStructured outputTemperatureAttachments
⚠ Providers report capability flags inconsistently
Modalities
TextImagePDF
Knowledge cutoff2025-08-31
Released / updated2026-02-05 / 2026-02-05

Quality & performanceArtificial Analysis · Intelligence Index v4.1 · rep. tier xhigh

Intelligence45.5
Value11
Coding
Agentic
Output speed125 tok/s
TTFT74.62 s
Value formulaIQ 45.5 ÷ min blended $4.331 = 11

Quality is independently evaluated by Artificial Analysis. Speed/latency are model-level medians.

Available at 27 providers27 with public prices

ProviderTierInputOutputCache readCache writeContextOutput limitStatus
RequestyGateway$1.57$12.60$0.158400,000128,000
PoeGateway$1.60$13.00$0.16400,000128,000
OrcaRouterGateway$1.75$14.00$0.175400,000128,000
NanoGPTGateway$1.75$14.00$0.175400,000128,000
ImpossiblGateway$1.75$14.00$0.175400,000128,000
OpenRouterGateway$1.75$14.00$0.175400,000128,000
OpenAIOfficialFirst-party$1.75$14.00$0.175400,000128,000
VivgridGateway$1.75$14.00$0.175400,000128,000
Cloudflare AI GatewayCloud$1.75$14.00$0.175400,000128,000
FastRouterGateway$1.75$14.00400,000128,000

Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.

Your usage cost

1Requesty$774.90
2Poe$797.20
3OrcaRouter$861.00
4NanoGPT$861.00
5Impossibl$861.00
6OpenRouter$861.00
7OpenAI · Official$861.00
Switch to Requesty to save $86.10/mo (10%). The gap is small, so staying on the official channel is fine.
Note: this is a gateway; verify availability and rate limits yourself.

Benchmark3 items

NameConditionsScoreMetricSource
SWE-Atlas Codebase QnAharness: Codex32.6scoreSource ↗
SWE-Atlas Refactoringharness: Codex42.38scoreSource ↗
SWE-Atlas Test Writingharness: Codex38.98scoreSource ↗

The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.

Reasoning control

effort = noneeffort = loweffort = mediumeffort = higheffort = maxbudget_tokenseffort = xhigheffort = minimal

3 / 27 providers expose no reasoning control (reasoning_options: []).

Related models

Price historyone sample accumulated per data sync

Input listOutput listMin blended
$14.00$02026-08-052026-08-212026-08-05 · Input list $1.752026-08-06 · Input list $1.752026-08-07 · Input list $1.752026-08-08 · Input list $1.752026-08-09 · Input list $1.752026-08-10 · Input list $1.752026-08-11 · Input list $1.752026-08-12 · Input list $1.752026-08-13 · Input list $1.752026-08-21 · Input list $1.752026-08-05 · Output list $14.002026-08-06 · Output list $14.002026-08-07 · Output list $14.002026-08-08 · Output list $14.002026-08-09 · Output list $14.002026-08-10 · Output list $14.002026-08-11 · Output list $14.002026-08-12 · Output list $14.002026-08-13 · Output list $14.002026-08-21 · Output list $14.002026-08-05 · Min blended $4.452026-08-06 · Min blended $4.452026-08-07 · Min blended $4.452026-08-08 · Min blended $4.452026-08-09 · Min blended $4.452026-08-10 · Min blended $4.452026-08-11 · Min blended $4.452026-08-12 · Min blended $4.452026-08-13 · Min blended $4.452026-08-21 · Min blended $4.33