GPT-4
Multimodal
Zero-eval
#1AI2 Reasoning Challenge (ARC)
#1Uniform Bar Exam
#1SAT Math
+3 more
by OpenAI
About
GPT-4 is a multimodal language model developed by OpenAI. It achieves strong performance with an average score of 77.7% across 12 benchmarks. It excels particularly in AI2 Reasoning Challenge (ARC) (96.3%), HellaSwag (95.3%), Uniform Bar Exam (90.0%). The model shows particular specialization in reasoning tasks with an average performance of 93.0%. The model is available through 2 API providers. As a multimodal model, it can process and understand text, images, and other input formats seamlessly.
Pricing Range
Input (per 1M)$30.00 -$30.00
Output (per 1M)$60.00 -$60.00
Providers2
Timeline
AnnouncedJun 13, 2023
ReleasedJun 13, 2023
Knowledge CutoffDec 31, 2022
Specifications
Capabilities
Multimodal
License & Family
License
Proprietary
Benchmark Performance Overview
Performance metrics and category breakdown
Overall Performance
12 benchmarks
Average Score
77.7%
Best Score
96.3%
High Performers (80%+)
8Performance Metrics
Max Context Window
65.5KAvg Throughput
102.0 tok/sAvg Latency
0msTop Categories
reasoning
93.0%
general
76.2%
math
68.5%
code
67.0%
Benchmark Performance
Top benchmark scores with normalized values (0-100%)
Ranking Across Benchmarks
Position relative to other models on each benchmark
AI2 Reasoning Challenge (ARC)
Rank #1 of 1
#1GPT-4
96.3%
HellaSwag
Rank #2 of 24
#1Claude 3 Opus
95.4%
#2GPT-4
95.3%
#3Gemini 1.5 Pro
93.3%
#4Claude 3 Sonnet
89.0%
#5Command R+
88.6%
Uniform Bar Exam
Rank #1 of 1
#1GPT-4
90.0%
SAT Math
Rank #1 of 1
#1GPT-4
89.0%
LSAT
Rank #1 of 1
#1GPT-4
88.0%
All Benchmark Results for GPT-4
Complete list of benchmark scores with detailed information
AI2 Reasoning Challenge (ARC) AI2 Reasoning Challenge (ARC) benchmark | reasoning | text | 0.96 | 96.3% | Self-reported |
HellaSwag HellaSwag benchmark | reasoning | text | 0.95 | 95.3% | Self-reported |
Uniform Bar Exam Uniform Bar Exam benchmark | general | text | 0.90 | 90.0% | Self-reported |
SAT Math SAT Math benchmark | math | text | 0.89 | 89.0% | Self-reported |
LSAT LSAT benchmark | general | text | 0.88 | 88.0% | Self-reported |
Winogrande Winogrande benchmark | reasoning | text | 0.88 | 87.5% | Self-reported |
MMLU MMLU benchmark | general | text | 0.86 | 86.4% | Self-reported |
DROP DROP benchmark | general | text | 0.81 | 80.9% | Self-reported |
MGSM MGSM benchmark | math | text | 0.74 | 74.5% | Self-reported |
HumanEval HumanEval benchmark | code | text | 0.67 | 67.0% | Self-reported |
Showing 1 to 10 of 12 benchmarks