o3
Multimodal
Zero-eval
#1ARC-AGI
#1MathVista
#1Tau-bench
+11 more
by OpenAI
About
o3 is a multimodal language model developed by OpenAI. It achieves strong performance with an average score of 65.3% across 19 benchmarks. It excels particularly in AIME 2024 (91.6%), ARC-AGI (88.0%), MathVista (86.8%). The model shows particular specialization in vision tasks with an average performance of 76.7%. It supports a 300K token context window for handling large documents. The model is available through 1 API provider. As a multimodal model, it can process and understand text, images, and other input formats seamlessly. Released in 2025, it represents OpenAI's latest advancement in AI technology.
Pricing Range
Input (per 1M)$2.00 -$2.00
Output (per 1M)$8.00 -$8.00
Providers1
Timeline
AnnouncedApr 16, 2025
ReleasedApr 16, 2025
Knowledge CutoffMay 31, 2024
Specifications
Capabilities
Multimodal
License & Family
License
Proprietary
Benchmark Performance Overview
Performance metrics and category breakdown
Overall Performance
19 benchmarks
Average Score
65.3%
Best Score
91.6%
High Performers (80%+)
8Performance Metrics
Max Context Window
300.0KAvg Throughput
50.0 tok/sAvg Latency
20msTop Categories
vision
76.7%
general
67.5%
agents
63.0%
math
51.3%
reasoning
47.3%
Benchmark Performance
Top benchmark scores with normalized values (0-100%)
Ranking Across Benchmarks
Position relative to other models on each benchmark
AIME 2024
Rank #5 of 41
#2Gemini 2.5 Pro
92.0%
#3Grok-3
93.3%
#4o4-mini
93.4%
#5o3
91.6%
#6DeepSeek-R1-0528
91.4%
#7Gemini 2.5 Flash
88.0%
#8o3-mini
87.3%
ARC-AGI
Rank #1 of 2
#1o3
88.0%
#2Qwen3-235B-A22B-Instruct-2507
41.8%
MathVista
Rank #1 of 35
#1o3
86.8%
#2o4-mini
84.3%
#3Kimi-k1.5
74.9%
#4Llama 4 Maverick
73.7%
AIME 2025
Rank #10 of 36
#7DeepSeek-R1-0528
87.5%
#8Gemini 2.5 Pro Preview 06-05
88.0%
#9Grok-3 Mini
90.8%
#10o3
86.4%
#11GPT-5 nano
85.2%
#12Gemini 2.5 Pro
83.0%
#13Qwen3 235B A22B
81.5%
VideoMMMU
Rank #3 of 3
#1Gemini 2.5 Pro Preview 06-05
83.6%
#2GPT-5
84.6%
#3o3
83.3%
All Benchmark Results for o3
Complete list of benchmark scores with detailed information
AIME 2024 AIME 2024 benchmark | general | text | 0.92 | 91.6% | Self-reported |
ARC-AGI ARC-AGI benchmark | reasoning | text | 0.88 | 88.0% | Self-reported |
MathVista MathVista benchmark | math | text | 0.87 | 86.8% | Self-reported |
AIME 2025 AIME 2025 benchmark | general | text | 0.86 | 86.4% | Self-reported |
VideoMMMU VideoMMMU benchmark | vision | multimodal | 0.83 | 83.3% | Self-reported |
GPQA GPQA benchmark | general | text | 0.83 | 83.3% | Self-reported |
MMMU MMMU benchmark | vision | multimodal | 0.83 | 82.9% | Self-reported |
Aider-Polyglot Aider-Polyglot benchmark | general | text | 0.81 | 81.3% | Self-reported |
CharXiv-R CharXiv-R benchmark | general | text | 0.79 | 78.6% | Self-reported |
MMMU-Pro MMMU-Pro benchmark | vision | multimodal | 0.76 | 76.4% | Self-reported |
Showing 1 to 10 of 19 benchmarks
Resources