Instructions to use FastFlowLM/GPT-OSS-20B-NPU2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use FastFlowLM/GPT-OSS-20B-NPU2 with Transformers:

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

pipe = pipeline("text-generation", model="FastFlowLM/GPT-OSS-20B-NPU2")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("FastFlowLM/GPT-OSS-20B-NPU2")
model = AutoModelForCausalLM.from_pretrained("FastFlowLM/GPT-OSS-20B-NPU2")
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

vLLM

How to use FastFlowLM/GPT-OSS-20B-NPU2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "FastFlowLM/GPT-OSS-20B-NPU2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FastFlowLM/GPT-OSS-20B-NPU2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/FastFlowLM/GPT-OSS-20B-NPU2

SGLang

How to use FastFlowLM/GPT-OSS-20B-NPU2 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 "FastFlowLM/GPT-OSS-20B-NPU2" \
    --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": "FastFlowLM/GPT-OSS-20B-NPU2",
		"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 "FastFlowLM/GPT-OSS-20B-NPU2" \
        --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": "FastFlowLM/GPT-OSS-20B-NPU2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use FastFlowLM/GPT-OSS-20B-NPU2 with Docker Model Runner:
```
docker model run hf.co/FastFlowLM/GPT-OSS-20B-NPU2
```

new_xclbin_v0.9.13

by FastFlowLM - opened Oct 8, 2025

base: refs/heads/main

←

from: refs/pr/1

Discussion Files changed

-10

Files changed (6) hide show

.gitattributes +0 -1
config.json +1 -1
dequant_mxfp4.xclbin +2 -2
dequant_q4_1.xclbin +2 -2
layer.xclbin +2 -2
mm.xclbin +2 -2

.gitattributes CHANGED Viewed

@@ -41,4 +41,3 @@ attn.xclbin filter=lfs diff=lfs merge=lfs -text
 dequant_mxfp4.xclbin filter=lfs diff=lfs merge=lfs -text
 expert.xclbin filter=lfs diff=lfs merge=lfs -text
 mm.xclbin filter=lfs diff=lfs merge=lfs -text
-*.q4nx filter=lfs diff=lfs merge=lfs -text

 dequant_mxfp4.xclbin filter=lfs diff=lfs merge=lfs -text
 expert.xclbin filter=lfs diff=lfs merge=lfs -text
 mm.xclbin filter=lfs diff=lfs merge=lfs -text

config.json CHANGED Viewed

@@ -78,5 +78,5 @@
   "addr_l_begin_mha": 20480,
   "addr_l_end_mha": 35328,
   "addr_kk": 35328,
-  "flm_version": "0.9.17"
 }

   "addr_l_begin_mha": 20480,
   "addr_l_end_mha": 35328,
   "addr_kk": 35328,
+  "flm_version": "0.9.12"
 }

dequant_mxfp4.xclbin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a7043d18b9aaefa0accea84c0fa730545cb4c944145b92f60100517f3be61247
-size 279435

 version https://git-lfs.github.com/spec/v1
+oid sha256:b52787eaf72fa34f5d540ee40c706ccc9a6c3f43a47190dea66cd5beadb84d8c
+size 280555

dequant_q4_1.xclbin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2bff000d64382748fbe4bece064b9442214b0a007ab5d5215c6e3e5453e3b31b
-size 114411

 version https://git-lfs.github.com/spec/v1
+oid sha256:18e24350c519521976c4a1fdb4f342a437ba9a4abedb716dca99adc5bb6e9a2a
+size 114667

layer.xclbin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b78eaf56dc153134fa2fe34313d9d8d71761f1c10d05b552cc2190f5defdf029
-size 452987

 version https://git-lfs.github.com/spec/v1
+oid sha256:828334475daecd108913fc83571b262c5420fb1882125aec0b2fa239cf6267ce
+size 445627

mm.xclbin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:99a9f2b47a09f2e129ee057f7b4b0e95955e7bfa1e48a99172bbd3992adc369a
-size 543675

 version https://git-lfs.github.com/spec/v1
+oid sha256:4d001fcc3e8b5aa553f740abedc859ded13173747c6eeb670763959322c8a936
+size 451611