Back to all models
L

Llama 3.1 405B

MetaReleased 2024-07-23
Official Page

Meta's largest open-weight model, rivaling GPT-4 class performance with full open weights for commercial and research use.

Overall Score
79%
Cost Efficiency
100
Context Window
128K

Strengths

  • ✓Fully open-source
  • ✓Massive 405B parameters
  • ✓Competitive with proprietary models
  • ✓Self-hostable

Weaknesses

  • ✗Requires significant infrastructure
  • ✗Text-only
  • ✗No official API
Architecture
Transformer (MoE)
Parameters
405B
Input
$0/M
Output
$0/M

Benchmark Performance

BenchmarkDescriptionScoreCategoryBar
MMLUMassive Multitask Language Understanding88.6%knowledge
HumanEvalCode generation accuracy89%coding
MATHMathematical reasoning90%math
GSM8KGrade school math93%math
GPQAGraduate-level science Q&A51.2%reasoning
HellaSwagCommon sense reasoning86.9%reasoning
SWE-benchSoftware engineering problems38.3%coding
ARC-CAI2 Reasoning Challenge95%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.