Danube3 4B
Open 4B small language model trained on 6 trillion tokens.
About
H2O-Danube3-4B is H2O.ai's open small language model, released July 18, 2024 as the flagship of the Danube3 series alongside the compact 500M variant. Trained on 6 trillion tokens, it scored over 80% on the 10-shot HellaSwag benchmark, which H2O.ai positioned as surpassing Apple's OpenELM-3B and rivaling Microsoft's Phi-3 4B. The series targets chat applications, research, and on-device and edge deployment where a 4B footprint fits.
Like H2O.ai's other releases, the weights are open source under Apache 2.0 and available on Hugging Face. The Danube line sits between H2O.ai's earlier Danube/Danube2 releases and remains a practical choice for efficient English-language deployment. Benchmark comparisons are vendor-reported.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 92; Trending score 74/100 from researcher assessment.