Head to head
Claude 3.5 Sonnet vs Llama 4 Behemoth (preview)
2.7×10²⁵Training compute (FLOP)5.2×10²⁵
$26MTraining cost$45M
Claude 3.5 Sonnet (Anthropic) and Llama 4 Behemoth (preview) (Meta AI) are both frontier AI models. Claude 3.5 Sonnet was published in June 2024 and Llama 4 Behemoth (preview) in April 2025.
Llama 4 Behemoth (preview) was trained on 5.2×10²⁵ FLOP, about 1.9x the compute of Claude 3.5 Sonnet at 2.7×10²⁵ FLOP. Training compute is the closest available proxy for how much was invested in a model, though it says nothing on its own about how well that compute was spent.
These two models share no benchmark on which both have been scored, so no direct performance comparison is possible here. The specification table below is a comparison of inputs, not of results.
Specifications
Anthropic
Organization
Meta AI
Jun 20, 2024
Published
Apr 5, 2025
2.7×10²⁵ FLOP
Training compute1.9x
5.2×10²⁵ FLOP
--
Parameters
2T
--
Dataset size
30T
--
Training hardware
NVIDIA H100 SXM5 80GB
--
Chips used
32,000
$26M
Training cost (2023 USD)1.7x
$45M
--
Training power draw
43.9 MW
API access
Accessibility
Unreleased
No
Open weights
No
United States
Country
United States
Related comparisons
Grok 4 vs Llama 4 Behemoth (preview)Grok 4 vs Claude 3.5 SonnetGPT-4.5 vs Llama 4 Behemoth (preview)GPT-4.5 vs Claude 3.5 SonnetGrok 3 vs Llama 4 Behemoth (preview)Grok 3 vs Claude 3.5 SonnetLlama 4 Behemoth (preview) vs Gemini 1.0 UltraLlama 4 Behemoth (preview) vs Llama Nemotron Ultra 253BLlama 4 Behemoth (preview) vs Grok-2Llama 4 Behemoth (preview) vs GPT-4 (Jun 2023)Llama 4 Behemoth (preview) vs Nemotron-4 340BLlama 4 Behemoth (preview) vs Inflection-2
SourceEpoch AI, 'AI Models'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/data/ai-models. Licensed under CC BY 4.0.