Instructions to use MiyaMusashi/Krono-Distill-20B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use MiyaMusashi/Krono-Distill-20B with Transformers:

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

pipe = pipeline("text-generation", model="MiyaMusashi/Krono-Distill-20B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("MiyaMusashi/Krono-Distill-20B", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use MiyaMusashi/Krono-Distill-20B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MiyaMusashi/Krono-Distill-20B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MiyaMusashi/Krono-Distill-20B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/MiyaMusashi/Krono-Distill-20B

SGLang

How to use MiyaMusashi/Krono-Distill-20B 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 "MiyaMusashi/Krono-Distill-20B" \
    --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": "MiyaMusashi/Krono-Distill-20B",
		"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 "MiyaMusashi/Krono-Distill-20B" \
        --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": "MiyaMusashi/Krono-Distill-20B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use MiyaMusashi/Krono-Distill-20B with Docker Model Runner:
```
docker model run hf.co/MiyaMusashi/Krono-Distill-20B
```

Krono-Distill-20B

Krono-Distill-20B is a derivative of openai/gpt-oss-20b, rebranded and specialized for:

strong Italian conversational quality
agentic workflows (tool use / function calling / multi-step planning)

This is not a from-scratch model. Weights descend from OpenAI GPT-OSS 20B (Apache-2.0).

Model details

Field	Value
Name	Krono-Distill-20B
Base model	`openai/gpt-oss-20b`
Type	Causal LM (derivative / post-trained)
License	Apache-2.0 (inherits base terms + attribution)
Intended use	General assistant, Italian-first, agent scaffolding
Author	MiyaMusashi

Provenance (required)

openai/gpt-oss-20b  →  Krono-Distill-20B

Current release stage:

L0: packaging + model card + lineage declared
L1 (current local): chat_template default model_identity → Krono-Distill (no longer ChatGPT)
L2+: identity/behavior baked via SFT (planned)

Identity policy

In normal chat, the model should present itself as Krono-Distill.
If asked explicitly about technical lineage, it may disclose that it is derived from gpt-oss-20b.

Usage (Transformers)

from transformers import pipeline

pipe = pipeline(
    "text-generation",
    model="MiyaMusashi/Krono-Distill-20B",
    torch_dtype="auto",
    device_map="auto",
)

messages = [
    {
        "role": "system",
        "content": "Sei Krono-Distill, un assistente forte sull'italiano e sulle capacità agentiche.",
    },
    {"role": "user", "content": "Chi sei e cosa sai fare bene?"},
]

# Prefer the model's chat template when available.
print(pipe(messages, max_new_tokens=256)[0]["generated_text"])

Limitations

May hallucinate facts; verify critical outputs.
Agentic quality depends on tool schemas and runtime scaffolding.
Italian strength improves with later SFT stages; early releases may still be close to the base.

Citation / attribution

@misc{krono-distill-20b,
  title  = {Krono-Distill-20B},
  author = {MiyaMusashi},
  year   = {2026},
  note   = {Derivative of openai/gpt-oss-20b},
  url    = {https://huggingface.co/MiyaMusashi/Krono-Distill-20B}
}

Please also credit OpenAI GPT-OSS:

https://huggingface.co/openai/gpt-oss-20b

Downloads last month: -; Downloads are not tracked for this model. How to track

Model tree for MiyaMusashi/Krono-Distill-20B

Base model

openai/gpt-oss-20b

Finetuned

(546)

this model