Sign Up

DeepReinforce

Reinforcement learning lab behind the Ornith coding models

ornith.aiUnited States

Follow
DeepReinforce is an AI research startup founded by Jiwei Li. It is known for the CUDA-L1 and CUDA-L2 reinforcement learning work on GPU code optimisation and for the open-source Ornith family of agentic coding models, released under the MIT license.

Catalogue

All models by DeepReinforce

5 models · 1–5
Select models to compareModelReleasedContextInput /MOutput /MParametersTypeI/OOpenness
Ornith 1.5 35B A3BDeepReinforceAugust 18, 2026262K——35BTexttext, image → textOpen weights
Ornith 1.0 31BDeepReinforceJune 25, 2026262K——31BTexttext, image → textOpen weights
Ornith 1.0 35BDeepReinforceJune 25, 2026262K——35BTexttext, image → textOpen weights
Ornith 1.0 397BDeepReinforceJune 25, 2026262K——397BTexttext, image → textOpen weights
Ornith 1.0 9BDeepReinforceJune 25, 2026262K——9BTexttext, image → textOpen weights
Page 1 of 1
PreviousNext

To give you the best experience, we use cookies for personalization, insights, and performance. Learn more.