Back to all models
G

GPT-4o

OpenAIReleased 2024-05-13
Official Page

OpenAI's flagship multimodal model capable of processing text, vision, and audio inputs. Excels at real-time voice conversations and multimodal reasoning.

Overall Score
79%
Cost Efficiency
100
Context Window
128K

Strengths

  • ✓Excellent multimodal capabilities
  • ✓Fast inference
  • ✓Strong coding performance
  • ✓Native audio processing

Weaknesses

  • ✗Limited context window compared to competitors
  • ✗Higher cost than some alternatives
Architecture
Transformer (MoE)
Parameters
Unknown
Input
$2.5/M
Output
$10/M

Benchmark Performance

BenchmarkDescriptionScoreCategoryBar
MMLUMassive Multitask Language Understanding88.7%knowledge
HumanEvalCode generation accuracy90.2%coding
MATHMathematical reasoning90%math
GSM8KGrade school math93%math
GPQAGraduate-level science Q&A49.9%reasoning
HellaSwagCommon sense reasoning87.3%reasoning
SWE-benchSoftware engineering problems38.7%coding
ARC-CAI2 Reasoning Challenge96%reasoning
AI
ModelBench

The open benchmark platform for comparing the latest AI models and their capabilities.

Platform

  • Compare Models
  • All Models
  • Methodology

AI Labs

  • OpenAI
  • Anthropic
  • Google DeepMind
  • Meta AI

Resources

  • Documentation
  • Benchmark Guide
  • API Reference
Benchmarks are sourced from official model cards and independent evaluations. Data is for informational purposes only.
© 2026 ModelBench. Built for the AI community.