Instructions to use tngtech/DeepSeek-R1T-Chimera with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tngtech/DeepSeek-R1T-Chimera with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="tngtech/DeepSeek-R1T-Chimera", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tngtech/DeepSeek-R1T-Chimera", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("tngtech/DeepSeek-R1T-Chimera", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use tngtech/DeepSeek-R1T-Chimera with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tngtech/DeepSeek-R1T-Chimera"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tngtech/DeepSeek-R1T-Chimera",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tngtech/DeepSeek-R1T-Chimera

SGLang

How to use tngtech/DeepSeek-R1T-Chimera with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "tngtech/DeepSeek-R1T-Chimera" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tngtech/DeepSeek-R1T-Chimera",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "tngtech/DeepSeek-R1T-Chimera" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tngtech/DeepSeek-R1T-Chimera",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tngtech/DeepSeek-R1T-Chimera with Docker Model Runner:
```
docker model run hf.co/tngtech/DeepSeek-R1T-Chimera
```

Can't find this model anywhere.

by lionyawn - opened Mar 8

Discussion

lionyawn

Mar 8

I'm not sure where else to put this, but I absolutely fell in love with this model and now there's no one that hosts it. It was pulled off OR and then Chutes completely. I know a lot of people that feel the same and miss it dearly. Are there any plans in trying to get this model hosted elsewhere in the future?

rmoriz

Mar 8

The successor https://huggingface.co/tngtech/DeepSeek-TNG-R1T2-Chimera
is available on Chutes via OpenRouter. They had to pull the free version to abuse or "artificially high demand" but it's cheap.
https://openrouter.ai/tngtech/deepseek-r1t2-chimera

lionyawn

Mar 8

The successor https://huggingface.co/tngtech/DeepSeek-TNG-R1T2-Chimera
is available on Chutes via OpenRouter. They had to pull the free version to abuse or "artificially high demand" but it's cheap.
https://openrouter.ai/tngtech/deepseek-r1t2-chimera

Hi, thanks for the response! I've tried R1T2 extensively and I still prefer R1T.

TNGHK

TNG Technology Consulting GmbH org Mar 9

Hello Lion,

thanks for your kind words. Can you send an email to info@tngtech.com (with a subject "for Henrik") please and I can respond to you?

Cheers,
HK

lionyawn

Mar 10

Hello Lion,

thanks for your kind words. Can you send an email to info@tngtech.com (with a subject "for Henrik") please and I can respond to you?

Cheers,
HK

Hi HK!

Of course! Email has been sent. 😄

Aelirenna

Mar 27

Same, I can't find it anywhere :( The most obedient model of all that I've used. Extremely upset(

TNGHK

TNG Technology Consulting GmbH org Mar 27

•

edited Mar 27

Did you try our experimental 1.3T model "The Artist"?
(PS: You can also send an email to info@tngtech.com and we can give you a test access)

almasedai

Apr 1

I'm having the exact same issue, trying to use other services than OR and chutes to get access to R1T... I'm on VoidAI right now but no cigar... It was my favorite hands down, and I don't like any other as much as I did R1T...

lionyawn

May 2

Did you try our experimental 1.3T model "The Artist"?
(PS: You can also send an email to info@tngtech.com and we can give you a test access)

Hi! I sent an email to this address and haven't heard back yet.

yeahmann

May 11

•

edited 3 days ago

I'm in the same boat! I prefer R1T and T2 even over the new V4 Pro. For me, it goes R1T>=R1T2>R1>R1 0528=V3 0324, with V4 Pro somewhere in the middle. R1T and R1T2 had my favorite personalities/styles of all the models I've used. Losing inference providers has been a real bummer.

Final edit: Chutes took down Chimera only a few days after I originally posted this comment. Made me feel like I willed it into existence. Hopefully someone else will host Chimera sometime, because they are still my favorites for writing, and I miss them. I don't know if The Artist is still in the works, or if it will be similar/an upgrade to Chimera. Part of me somehow hopes it's a four way merge between the components of R1T2 along with V4's ability to handle long context, or that you guys have found another way to handle the problem of newer, more intelligent models being worse (or at least less exciting) writers. Maybe that's delusional on my part, but I'm still looking forward to anything you guys make. If nothing else, thanks for the good times and the happy memories.

loherj

May 15

This comment has been hidden

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment