0

OpenThinker (series)

Open weights

Open Thoughts' fully-open reasoning model series - Qwen2.5-base SFT on R1-distilled traces, with 7B/32B variants across three generations.

Publisher
Open Thoughts
Params
7B / 32B
Context
16K
Input
Output
License
Apache-2.0
Released
Jan 2025

Cite

Notes

Only stored in your browser.

Context
16K

Reported eval scores

0
No public eval scores tracked.

Introduced in

paperOpen Thoughts: Curating Reasoning Datasets for Open-Source R1 Replications