Model

GPT-5.4 Pro

CLOSED

OpenAI · GPT-5.4 family · released Mar 5, 2026

Top-performance tier of the GPT-5.4 family, launched same-day as the flagship/Thinking tier (Mar 5, 2026); available on ChatGPT Pro/Enterprise plans and via the Responses API only (not Chat Completions). Priced at $30/$180 per M tokens on a shared 1.05M-token context window. Interestingly, on GDPval the standard Thinking tier (83.0%) actually beats Pro (82.0%) -- OpenAI's own chart flags this explicitly.

ReasoningCodingVisionFunction callingTool useAgentic

2786.0

Elo · rank #5

Parameters

Undisclosed

Active params

—

Context

1.05M tokens

Architecture

Undisclosed architecture (presumed dense transformer, reasoning-tuned), same family as GPT-5.4 flagship but run at maximum reasoning effort ('Highest' reasoning / 'Slowest' speed per OpenAI's own model card); Responses-API-only, supports multi-turn background processing for long-running tasks

License

Proprietary (OpenAI API Terms)

Languages

—

API price (in/out)

$30 / $180

Modalities

text · vision

Benchmark results

Bar shows position within the tracked field; marker = field best

ARC-AGI-1Reasoning94.5%#2

best: GPT-5.6 · 96.5%

ARC-AGI-2Reasoning83.3%#3

best: GPT-5.6 · 92.5%

BrowseCompAgents89.3%#3

best: Kimi K3 · 91.2%

FrontierMath (Tiers 1-3)Math50.0%#5

best: GPT-5.6 · 89.0%

FrontierMath (Tier 4)Math38.0%#2

best: GPT-5.6 · 83.0%

GDPvalAgents82.0%#4

best: Seed 2.1 Pro · 87.9%

GPQA DiamondReasoning94.4%#2

best: GPT-5.6 · 94.6%

Humanity's Last ExamReasoning42.7%#20

best: Claude Sonnet 5 · 57.4%

Run it locally

Closed weights — available via API only. No local deployment.

Input / M tok

$30

Output / M tok

$180

GPT-5.4 family

Elo progression across releases

GPT-5.4Mar 20262473.4 GPT-5.4 ProMar 20262786.0 GPT-5.4 miniMar 20262098.8 GPT-5.4 nanoMar 20261965.6

Resources

DOCShttps://developers.openai.com/api/docs/models/gpt-5.4-pro APIAvailable via OpenAI API

API price $30/$180 · each benchmark row carries its own source badge (see methodology)