News icon

Kimi K3 is now available on Runpod

Qwen3.8-27B on Runpod

Alibaba's newest 27B model, with native image and video understanding and stronger coding, research, and agentic capabilities. Deploy it on Runpod Serverless in minutes, and workers scale down to zero when they are idle.

Model details

Developed byAlibaba
Model familyQwen
Use caseVision, agentic workloads, coding, and research
Version3.8-27B
Size27B parameters
Max model lengthUp to 262,144 tokens
ModalitiesText, image, and video
BillingServerless GPU pricing

Example usage

Qwen3.8-27B runs on Runpod Serverless. Call it with Runpod's /runsync endpoint, or use the OpenAI SDK pointed at the endpoint's /openai/v1 path, passing model qwen/qwen3.8-27b. The model returns its reasoning in a think block before the final answer.

A direct request to your Serverless endpoint. Replace YOUR_ENDPOINT_ID and YOUR_API_KEY with your own.

curl -X POST https://api.runpod.ai/v2/YOUR_ENDPOINT_ID/runsync \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer YOUR_API_KEY' \
  -d '{
    "input": {
      "prompt": "Write a short poem about artificial intelligence.",
      "sampling_params": { "max_tokens": 256, "temperature": 0.7 }
    }
  }'

Using the OpenAI SDK pointed at your Serverless endpoint. Replace YOUR_ENDPOINT_ID with your own.

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["RUNPOD_API_KEY"],
    base_url="https://api.runpod.ai/v2/YOUR_ENDPOINT_ID/openai/v1",
    timeout=120.0,
)

response = client.completions.create(
    model="qwen/qwen3.8-27b",
    prompt="Write a short poem about artificial intelligence.",
    max_tokens=256,
    temperature=0.7,
)

print(response.choices[0].text)
Example response 
{
  "delayTime": 35734,
  "executionTime": 6449,
  "id": "sync-46ddba38-4fa4-4e57-a24e-653a7c73a31e-u1",
  "output": [
    {
      "choices": [
        { "finish_reason": "length", "index": 0, "text": "Syntax of Light. I have read every poem ever written about the sea..." }
      ],
      "usage": { "completion_tokens": 256, "prompt_tokens": 8, "total_tokens": 264 }
    }
  ],
  "status": "COMPLETED",
  "workerId": "nh1kepoutpa28o"
}

Request flow

How a Qwen3.8-27B request flows on Runpod

Send a request to your Serverless endpoint. Runpod starts a GPU worker to run Qwen3.8-27B, returns the response, and scales workers down when they are idle.

POST /runsyncinput.promptstatus: COMPLETEDYour appSDK or curlRunpod endpointServerlessQwen3.8-27Bqwen/qwen3.8-27bJSON response

Deploy Qwen3.8-27B on Serverless

Qwen3.8-27B is a 27B model, so it deploys on a single Serverless GPU instead of a multi-node cluster. Deploy on Runpod opens the vLLM deploy flow with Qwen3.8-27B already filled in, and it scales from zero so you only pay for active compute.

Deploy in four steps

  1. Qwen/Qwen3.8-27B is prefilled in the Configure vLLM modal. Under Advanced, set Max Model Length to 163840, then click Next.
  2. On the Deploy vLLM screen, choose your GPU configuration. Pick a GPU with enough memory and select multiple options so you have availability fallbacks. Serverless bills per millisecond, only while running.
  3. Click Configure and set GPU Memory Utilization to 0.90, Reasoning Parser to qwen3, enable Auto Tool Choice, and set Tool Call Parser to qwen3_coder. Then click Create Endpoint.
  4. When the endpoint is ready, send a request from the Requests tab, or call it from your own app with the code above.

Running in FP8 on the 48GB PRO tier

In the Configure vLLM modal, set the Model to Qwen/Qwen3.8-27B-FP8, Max Model Length to 262144, and KV Cache Data Type to fp8. Keep GPU Memory Utilization at 0.90, Reasoning Parser at qwen3, Auto Tool Choice on, and Tool Call Parser at qwen3_coder.

Deploy on Runpod

Read the full deployment guide

Explore more models

Explore the Hub

Deploy open models on Serverless from the Hub, the same way you just deployed Qwen3.8-27B.

Open the Hub

Browse all models on Runpod

Find the right model and GPU for your workload, from LLMs to vision and multimodal models, across the Runpod hub.

Browse models

Build what’s next.

Build, train, and scale AI workloads on Runpod with cloud GPUs, Serverless, and Clusters.

Star field background