OpenHermes 2.5 Mistral 7B
Teknium's million example Mistral fine tune that powered Hermes 2.
About
OpenHermes 2.5 Mistral 7B is Teknium's March 2024 continuation of the OpenHermes line, fine tuning Mistral 7B on roughly one million entries of primarily GPT-4 generated data plus high quality open datasets, with a notable share of code instruction data. It improved on OpenHermes 2 across GPT4All, AGIEval, TruthfulQA, and HumanEval.
The dataset became the foundation for Nous Research's Hermes 2 family, which trained its flagship Mixtral DPO model on the OpenHermes 2.5 collection. Released under Apache 2.0, it remains a reference point for the 2024 wave of synthetic data fine tunes.
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 122; Trending score 72/100 from researcher assessment.