GP

GPT-5.4 mini

Reasoning modelGPT-5.4

GPT-5.4 mini

Release date: 2026-03-17Updated: 2026-06-15 07:18:17.482Knowledge cutoff: 2025-08-311,181
Live demoGitHubHugging FaceCompare
Parameters
Not disclosed
Context length
400K
Chinese support
Supported
Reasoning ability

GPT-5.4 mini is an AI model published by OpenAI, released on 2026-03-17, for Reasoning model, and 400K context length, under the 不开源 license, with a 93.40 score on τ²-Bench - Telecom.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

GPT-5.4 mini

Model basics

Reasoning traces
Supported
Thinking modes
Thinking Level · High (Default)Thinking Level · LowThinking Level · MediumThinking Level · Extra-High
Context length
400K tokens
Max output length
128K tokens
Model type
Reasoning model
Modality (in / out)
Text, Image → Text
Release date
2026-03-17
Model file size
No data
MoE architecture
No
Total params / Active params
No data / N/A
Knowledge cutoff
2025-08-31
GPT-5.4 mini

Open source & experience

Code license
不开源
Weights license
不开源
GitHub repo
GitHub link unavailable
Hugging Face
Hugging Face link unavailable
GPT-5.4 mini

Official resources

DataLearnerAI blog
No blog post yet
GPT-5.4 mini

API details

API speed
4/5
💡Default unit: $/1M tokens. If vendors use other units, follow their published pricing.
Standard
TypeConditionInputOutput
Text-$0.750/ 1M$4.50/ 1M
Batch
TypeConditionInputOutput
Text-$0.375/ 1M$2.25/ 1M
TextCache = 5 min$0.038/ 1M$2.25/ 1M
Cache PricingPrompt Cache
TypeTTLWriteRead
Text5m$0.075/ 1M$4.50/ 1M
GPT-5.4 mini

Benchmark Results

GPT-5.4 mini currently shows benchmark results led by GPQA Diamond (33 / 179, score 88), Tool Decathlon (2 / 7, score 42.90), HLE (47 / 159, score 41.50). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.

Thinking
Tool usage

General Knowledge

3 evaluations
Benchmark / mode
Score
Rank/total
GPQA Diamond
Extra-High
88
33 / 179
HLE
Extra-High
28.20
83 / 159
HLE
Extra-HighTools
41.50
47 / 159

Math and Reasoning

1 evaluations
Benchmark / mode
Score
Rank/total
2.10
56 / 80

Coding and Software Engineer

1 evaluations
Benchmark / mode
Score
Rank/total
SWE-Bench Pro - Public
Extra-HighTools
54.40
22 / 44

Agent Level Benchmark

1 evaluations
Benchmark / mode
Score
Rank/total
τ²-Bench - Telecom
Extra-HighTools
93.40
17 / 35

AI Agent - Tool Usage

3 evaluations
Benchmark / mode
Score
Rank/total
OSWorld-Verified
Extra-HighTools
72.10
11 / 18
Terminal Bench 2.0
Extra-HighTools
60
19 / 46
Tool Decathlon
Extra-HighTools
42.90
2 / 7

Claw-style Agent Evaluation

1 evaluations
Benchmark / mode
Score
Rank/total
Claw Bench
Thinking ModeTools
75.30
25 / 29

Compare with other models

GPT-5.4 mini

Publisher

GPT-5.4 mini

Model Overview

GPT-5.4 mini is an AI model published by OpenAI, released on 2026-03-17, for Reasoning model, and 400K context length, under the 不开源 license, with a 93.40 score on τ²-Bench - Telecom.

DataLearner on WeChat

Follow DataLearner on WeChat for AI model updates and research notes.

DataLearner WeChat QR code