ModelBeat
Models/GPT-5.5
All models
O

OpenAI: GPT-5.5

openai/gpt-5.5

GPT-5.5 is OpenAI’s frontier model designed for complex professional workloads, building on GPT-5.4 with stronger reasoning, higher reliability, and improved token efficiency on hard tasks.

Context
1.1M
Input
$5 / 1M
Output
$30 / 1M
Released
Apr 23, 2026
Modalities
fileimagetexttext

Benchmarks

Scores on standardized evaluations. Higher is better — and rank shows where GPT-5.5 lands among all models tracked on Model Beat.

93.0
Intelligence Index
Epoch AI
93rd percentile of tracked models
89.0
Coding Index
Epoch AI
89th percentile of tracked models
95.0
Agentic Index
Epoch AI
95th percentile of tracked models

Reasoning

6 evals
ARC-AGI95.0%

Abstract visual reasoning

ARC-AGI-285.0%

Harder abstract reasoning

GPQA Diamond94.0%

Graduate-level scientific reasoning

SimpleBench69.0%

Common-sense trick questions

SimpleQA Verified63.1%

Factual accuracy & hallucination

WeirdML84.9%

Novel ML problem-solving

Coding

3 evals
GSO (code optimization)40.2%

Code performance optimization

SWE-bench Verified80.6%

Real GitHub issue resolution

WebDev Arena1505

Human-rated web development

Math

3 evals
AIME 2024/2025100.0%

Olympiad-qualifier math

FrontierMath51.7%

Research-level math problems

FrontierMath Tier 435.4%

Hardest research math

Agentic & Tools

2 evals
APEX38.4%

Multi-step agentic tasks

Terminal-Bench84.7%

Command-line agentic tasks

In the news

13

Model & benchmark data from Epoch AI (CC BY); pricing, specs & descriptions from OpenRouter.