Model

GPT-5.4 nano

CLOSED

OpenAI · GPT-5.4 family · released Mar 17, 2026

Cheapest tier of the GPT-5.4 family, launched Mar 17 2026 alongside GPT-5.4 mini via OpenAI's own announcement as a distinct, separately-priced ($0.20/$1.25 vs GPT-5 nano's $0.05 per-M input per OpenAI's own pricing-comparison table), separately-benchmarked model -- not a repricing of GPT-5 nano. API-only (no ChatGPT/Codex availability); OpenAI's own model card lists computer-use as 'Not supported' for this tier (unlike mini/flagship).

ReasoningCodingVisionFunction callingTool useAgentic

1929.3

Elo · rank #82

Parameters

Undisclosed

Active params

—

Context

400K tokens

Architecture

Undisclosed architecture (presumed dense transformer); cheapest/fastest tier of the GPT-5.4 family, targeted at classification, extraction, ranking, and lightweight coding subagents; lacks native computer-use tool support that mini/flagship have

License

Proprietary (OpenAI API Terms)

Languages

—

API price (in/out)

$0.2 / $1.25

Modalities

text · vision

Benchmark results

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

AIMEMath87.8%#55

best: GPT-5.2 · 100.0%

GDPval-AAAgents1100#41

best: Claude Fable 5 · 1932

GPQA DiamondReasoning82.8%#74

best: GPT-6 Astra · 96.0%

Humanity's Last ExamReasoning24.3%#64

best: Claude Opus 5 · 64.7%

IFBenchReasoning76.0%#18

best: MiniMax M3 · 83.0%

LiveCodeBenchCoding84.0%#30

best: DeepSeek-V4-Pro (Think Max) · 93.5%

MMMU-ProVision66.1%#41

best: Claude Opus 4.7 · 85.5%

OSWorld-VerifiedAgents39.0%#29

best: Claude Fable 5 · 85.0%

SWE-bench ProCoding52.4%#43

best: Claude Fable 5.1 · 81.2%

SWE-bench VerifiedCoding69.8%#63

best: Claude Opus 5 · 96.0%

τ²-Bench TelecomAgents92.5%#18

best: Claude Opus 4.6 · 99.3%

Terminal-Bench 2.0Coding46.3%#66

best: Gemini 3.8 Flash · 89.4%

Run it locally

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

Input / M tok

$0.2

Output / M tok

$1.25

GPT-5.4 family

Elo progression across releases

GPT-5.4Mar 20262445.3 GPT-5.4 ProMar 20262759.5 GPT-5.4 miniMar 20262065.9 GPT-5.4 nanoMar 20261929.3

Resources

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

API price $0.2/$1.25 · each benchmark row carries its own source badge (see methodology)