DeepSeek-R1T-Chimera

Model merge of DeepSeek-R1 and DeepSeek-V3 (0324)

An open weights model combining the intelligence of R1 with the token efficiency of V3.

For details on the construction process and analyses of Chimera model variants, please read our paper.

Paper on arXiV | Announcement on X | LinkedIn post | Try it on OpenRouter

Update: we released R1T2-Chimera that is both faster and smarter than R1.

Model Details

• Architecture: DeepSeek-MoE Transformer-based language model
• Combination Method: Merged model weights from DeepSeek-R1 and DeepSeek-V3 (0324)
• Release Date: 2025-04-27

Use, Out-of-scope Use, Limitations, Risks, Recommendations et al

Regarding R1T Chimera, we ask you to follow the careful guidelines that Microsoft has created for their "MAI-DS-R1" DeepSeek-based model.

These guidelines are available here on Hugging Face.

Contact

• Email: research@tngtech.com
• X.com: @tngtech

Citation

@misc{tng_technology_consulting_gmbh_2025,
	author       = { TNG Technology Consulting GmbH },
	title        = { DeepSeek-R1T-Chimera },
	year         = 2025,
    month        = {April},
	url          = { https://huggingface.co/tngtech/DeepSeek-R1T-Chimera },
	doi          = { 10.57967/hf/5330 },
	publisher    = { Hugging Face }
}