Model · Text generation
Llama-3.1-8B-Instruct
Published by Meta Llama
Why it is here: The instruction-tuned 8B that most fine-tunes start from.
Collection · 6 entries
Open-weight text models worth knowing
Widely used open-weight language models, chosen because each one is a distinct family rather than a variant of the one above it. Selection, not a ranking.
Edited by SAVRN research · last revised 2026-09-18
Entries in this collection
Model · Text generation
Mistral-7B-Instruct-v0.2
Published by Mistral AI_
Why it is here: A small dense model with a long record of deployment.
Model · Text generation
Qwen2.5-7B-Instruct
Published by Qwen
Why it is here: The Qwen family's general-purpose instruct release.
Model · Text generation
DeepSeek-R1
Published by DeepSeek
Why it is here: A reasoning-trained release published with open weights.
Model · Text generation
gemma-2-9b-it
Published by Google
Why it is here: Google's open-weight instruct model.
Model · Text generation
gpt-oss-120b
Published by OpenAI
Why it is here: An open-weight release priced by several hosts on the SAVRN Index.
Explore More
Every collection · The model directory · How a collection is decided