OpenAI

GPT-5

Multimodal
#1MMMU

by OpenAI

+
+
+
+
About

GPT-5, released by OpenAI on August 7, 2025, is a large language model that combines direct generation and extended reasoning in a single unified system with a built-in routing mechanism. It features a 400K token context window, 128K maximum output tokens, native multimodal support (text, image, audio, video), and demonstrated strong results across coding, mathematics, visual understanding, and health benchmarks at release. GPT-5 targets complex multi-step tasks including advanced coding, mathematical problem solving, and long-context analysis.

+
+
+
+
Pricing Range
Input (per 1M)$30.00 -$30.00
Output (per 1M)$120.00 -$120.00
Providers1
+
+
+
+
Timeline
ReleasedAug 7, 2025
Knowledge CutoffMay 1, 2025
+
+
+
+
Specifications
Capabilities
Multimodal
+
+
+
+
License & Family
License
Proprietary
Performance Overview
Performance metrics and category breakdown

Overall Performance

8 benchmarks
Average Score
55.7%
Best Score
87.3%
High Performers (80%+)
2

Performance Metrics

Max Context Window
432.8K

Top Categories

Science
87.3%
Multimodal
78.4%
Coding
74.9%
Agents
45.0%
Tool Use
44.5%
+
+
+
+
All Benchmark Results for GPT-5
Complete list of benchmark scores with detailed information
GPQA Diamond
Science
87.30
87.3%
Unverified
τ-bench
Agents
80.00
80.0%
Unverified
MMMU
Multimodal
78.40
78.4%
Unverified
SWE Bench Verified
Coding
74.90
74.9%
Unverified
MCP-Atlas
Tool Use
44.50
44.5%
Unverified
GDPVal
Agents
34.80
34.8%
Unverified
Humanity's Last Exam
Reasoning
25.32
25.3%
Unverified
BrowseComp
Agents
20.10
20.1%
Unverified