H2O-Danube3-500M
514M-parameter small LLM that runs fully offline on modern smartphones.
About
H2O-Danube3-500M is a chat fine-tuned small language model from H2O.ai, released on July 18, 2024 as part of the Danube3 series (500M and 4B). It adjusts the Llama 2 architecture to around 500 million parameters with an 8192-token context window, trained with H2O LLM Studio.
It is designed for efficient inference on consumer hardware and edge devices, and can run fully offline on modern smartphones via H2O AI Personal GPT. All Danube3 models are openly available under the Apache 2.0 license on Hugging Face.
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 — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 123; Trending score 60/100 from researcher assessment.