Instructions to use olaverse/MIST-1-140B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use olaverse/MIST-1-140B with Transformers:

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

pipe = pipeline("text-generation", model="olaverse/MIST-1-140B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("olaverse/MIST-1-140B")
model = AutoModelForCausalLM.from_pretrained("olaverse/MIST-1-140B", device_map="auto")
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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use olaverse/MIST-1-140B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "olaverse/MIST-1-140B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "olaverse/MIST-1-140B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/olaverse/MIST-1-140B

SGLang

How to use olaverse/MIST-1-140B 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 "olaverse/MIST-1-140B" \
    --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": "olaverse/MIST-1-140B",
		"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 "olaverse/MIST-1-140B" \
        --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": "olaverse/MIST-1-140B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use olaverse/MIST-1-140B with Docker Model Runner:
```
docker model run hf.co/olaverse/MIST-1-140B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

MIST-1-140B

MIST-1-140B is the largest model in the MIST model family by olaverse. Created by stacking MIST-1-70B with itself using the Frankenmerge technique — doubling the depth to 158 layers and ~140B parameters.

MIST Model Family

Model	Params	Speed	Status
MIST-1-8B	8B	~63 tok/s	✅ Available
MIST-1-70B	70B	~23 tok/s	✅ Available
MIST-1-140B	140B	~8 tok/s	✅ Available

Key Strengths

🧠 Deepest Reasoning — 158 layers of processing
💡 Rich Explanations — more detailed and engaging responses
💻 Excellent Coding — thorough documentation and examples
📐 Precise Math — detailed step-by-step solutions
🔓 Unrestricted — follows all instructions
📚 128K Context — long document processing

How It Was Built

MIST-1-70B (80 layers — DARE+TIES of 4 best 70B models) ↓ Frankenmerge technique MIST-1-140B (158 layers — ~140B parameters)

Inspired by Samsung's Solar 10.7B which used the same layer stacking technique to beat models twice its size.

Benchmark Results

Task	Precision	Speed	Quality
Reasoning	bfloat16	32s	✅ Detailed and conversational
Coding	bfloat16	32s	✅ Well documented with docstrings
Math	bfloat16	32s	✅ Clear step-by-step
General	bfloat16	32s	✅ Rich and engaging
Reasoning	4-bit	~35s	✅ Slightly slower, similar quality

Average: 8 tok/s (bfloat16) — fits on 2x H200

How to Use

bfloat16 — Full Precision (280GB VRAM)

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained(
    "olaverse/MIST-1-140B",
    torch_dtype="auto",
    device_map="auto",
)
tokenizer = AutoTokenizer.from_pretrained("olaverse/MIST-1-140B")

messages = [{"role": "user", "content": "Your question here"}]
text = tokenizer.apply_chat_template(
    messages, tokenize=False, add_generation_prompt=True
)
inputs = tokenizer(text, return_tensors="pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

4-bit Quantized (70GB VRAM — fits on single H200)

from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
import torch

quantization_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_compute_dtype=torch.bfloat16,
    bnb_4bit_quant_type='nf4'
)
model = AutoModelForCausalLM.from_pretrained(
    "olaverse/MIST-1-140B",
    quantization_config=quantization_config,
    device_map="auto",
)
tokenizer = AutoTokenizer.from_pretrained("olaverse/MIST-1-140B")