OpenAI logo

GPT-5.5

Multimodal
OpenAI

GPT-5.5 is OpenAI's smartest model yet, designed for real work across agentic coding, computer use, knowledge work, and early scientific research. It matches GPT-5.4 per-token latency in real-world serving while reaching a much higher level of intelligence and using significantly fewer tokens to complete the same tasks, with a 1M-token context window in the API.

Key Specifications

Parameters
-
Context
1.1M
Release Date
April 23, 2026
Average Score
66.4%

Timeline

Key dates in the model's history
Announcement
April 23, 2026
Today
August 27, 2026

Technical Specifications

Parameters
-
Training Tokens
-
Knowledge Cutoff
December 1, 2025
Family
-
Capabilities
MultimodalZeroEval

Pricing & Availability

Input (per 1M tokens)
$5.00
Output (per 1M tokens)
$30.00
Max Input Tokens
1.1M
Max Output Tokens
128.0K
Supported Features
Function CallingStructured OutputCode ExecutionWeb SearchBatch InferenceFine-tuning

Benchmark Results

Model performance metrics across various tests and benchmarks

Reasoning

Logical reasoning and analysis
GPQA
GPQA Diamond. Reasoning effort xhigh.Self-reported
93.6%

Other Tests

Specialized benchmarks
ARC-AGI
ARC-AGI-1 (Verified). Reasoning effort xhigh.Self-reported
95.0%
ARC-AGI v2
ARC-AGI-2 (Verified). Reasoning effort xhigh.Self-reported
85.0%
BixBench
BixBench bioinformatics data analysis. Reasoning effort xhigh.Self-reported
80.5%
BrowseComp
BrowseComp agentic web browsing benchmark. Reasoning effort xhigh.Self-reported
84.4%
CyberGym
CyberGym. Reasoning effort xhigh.Self-reported
81.8%
DeepSWE 1.1
DeepSWE v1.1 leaderboard, mini-swe-agent harness, xhigh effort; Pass@1 67% ± 6%.Self-reported
67.0%
Finance Agent
FinanceAgent v1.1. Reasoning effort xhigh.Self-reported
60.0%
Finance Agent v2
Self-reported
51.8%
FrontierCode 1.1
FrontierCode 1.1 current leaderboard; mergeability score at xhigh effort: 43.0%.Self-reported
43.0%
FrontierMath
FrontierMath Tier 4. Reasoning effort xhigh.Self-reported
35.4%
FrontierSWE
CodexSelf-reported
73.0%
GDPval-MM
GDPval (wins or ties). Reasoning effort xhigh.Self-reported
84.9%
GeneBench
GeneBench. Reasoning effort xhigh.Self-reported
25.0%
Graphwalks BFS >128k
Graphwalks BFS 1M f1. Reasoning effort xhigh.Self-reported
45.4%
Graphwalks parents >128k
Graphwalks parents 1M f1. Reasoning effort xhigh.Self-reported
58.5%
Humanity's Last Exam
Humanity's Last Exam (with tools). Reasoning effort xhigh.Self-reported
52.2%
Legal Agent Benchmark
All-pass rate (Harvey held-out set)Self-reported
2.1%
LiveBench
2026-01-08, Thinking xHigh EffortSelf-reported
80.7%
MCP Atlas
MCP Atlas (Scale AI April 2026 update). Reasoning effort xhigh.Self-reported
75.3%
MMMU-Pro
MMMU Pro (with tools). Reasoning effort xhigh.Self-reported
83.2%
MRCR v2 (8-needle)
OpenAI MRCR v2 8-needle, 512K-1M context range. Reasoning effort xhigh.Self-reported
74.0%
OfficeQA Pro
OfficeQA Pro. Reasoning effort xhigh.Self-reported
54.1%
OSWorld-Verified
OSWorld-Verified desktop navigation benchmark. Reasoning effort xhigh.Self-reported
78.7%
SWE-Bench Pro
SWE-Bench Pro (Public). Reasoning effort xhigh.Self-reported
58.6%
Tau2 Telecom
Tau2-bench Telecom (original prompts, no prompt tuning). Reasoning effort xhigh.Self-reported
98.0%
Terminal-Bench 2.0
Terminal-Bench 2.0. Reasoning effort xhigh.Self-reported
82.7%
Toolathlon
Toolathlon tool-use agentic tasks benchmark. Reasoning effort xhigh.Self-reported
55.6%

License & Metadata

License
proprietary
Announcement Date
April 23, 2026
Last Updated
August 27, 2026

Similar Models

All Models

Recommendations are based on similarity of characteristics: developer organization, multimodality, parameter size, and benchmark performance. Choose a model to compare or go to the full catalog to browse all available AI models.