Training data
The teacher was Qwen/Qwen3.8-Flash-Next-FP8, running the vanillux agent.
The training set contains 437 completed, text-only trajectories with a
family-specific verifier score of at least 0.8:
Table with columns: Task family, Trajectories, Distinct tasks| Task family | Trajectories | Distinct tasks |
|---|
| Clinical trial matching | 116 | 15 |
| EHR data quality | 321 | 36 |
| Total | 437 | 51 |
The data comes from the 20260904-flash-next-medcli-k8 and
20260906-flash-next-medcli-k8 collection runs, exported as
medcli-text-score0.8.jsonl (approximately 69.3 million context and response
tokens). Training uses the captured token IDs and assistant loss masks,
preserving the teacher's context without reapplying a chat template.
Training configuration
Table with columns: Setting, Value| Setting | Value |
|---|
| Base model | Qwen/Qwen3.8-27B |
| Recipe | Supervised fine-tuning with Slime and Megatron-LM |
| Epochs / completed optimizer steps | 1 configured epoch / 53 steps |
| Batch size | 8 trajectories per step |
| Learning rate | 1e-5, cosine decay to 1e-6, 10% warmup |
| Weight decay / Adam beta2 / gradient clipping | 0.1 / 0.95 / 1.0 |
| Maximum tokens per micro-batch | 235,000 |
| Parallelism | 8 GPUs; tensor parallelism 4, pipeline parallelism 2 |
| Memory configuration | Full activation recomputation and optimizer CPU offload |
| Checkpoint | iter_0000053-complete |
The run's env.sh and README.md record the configuration and outcomes.
The data manifest is at
examples/runs/20260908-medcli-sft-qwen3.5-4b-q38/data/medcli-text-score0.8.manifest.json;
this run uses the same exported training set.
HealthAgentBench evaluation: 100-turn budget
The complete checkpoint was evaluated with vanillux 0.4.2, a maximum of
100 agent steps per trial, and three attempts on each of 54 tasks.
The table reports mean reward from the clean complete-checkpoint rerun
recorded in the training run's README.
Table with columns: Task family, Base Qwen3.8-27B, This checkpoint| Task family | Base Qwen3.8-27B | This checkpoint |
|---|
| Clinical trial matching | 0.333 | 0.481 |
| EHR data quality | 0.167 | 0.417 |
| EHR-to-MEDS ETL | 1.000 | 1.000 |
| X-ray report correction | 0.233 | 0.200 |
| CT abnormality | 0.000 | 0.033 |
| Pathology tumor selection | 0.067 | 0.000 |
| EHR event modeling | 0.500 | 0.333 |
| Overall | 0.206 | 0.235 |
The base results are from the September 3 evaluation under the same benchmark
protocol. These are small task sets with three attempts per task; results vary
across families. The run README also records a separate 512-step evaluation.
Checkpoint contents
Verified weight contents: All 333 vision tensors are present in
model-restored-frozen.safetensors (approximately 879 MiB). No MTP weights
are present: all 15 MTP tensors are omitted, although config.json still sets
text_config.mtp_num_hidden_layers to 1. These contents were checked against
the actual safetensors headers and the weight index.
This directory contains the trained language weights and the base model's
333 frozen vision tensors restored after the initial HF export, for 1,184
indexed tensors in total. The 15 multi-token prediction (MTP) tensors remain
omitted; RESTORED.json records the restoration and omissions. The tokenizer,
chat template, and image/video processor files are inherited from the base model.
Original Qwen3.8-27B model card
The original model card is preserved below. Its model description, benchmarks,
and usage examples refer to the upstream base model.
Qwen3.8-27B
[!Note]
This repository contains model weights and configuration files for the post-trained model in the Hugging Face Transformers format.
These artifacts are compatible with Hugging Face Transformers, vLLM, SGLang, TokenSpeed, etc.
[!Tip]
For users seeking managed, scalable inference without infrastructure maintenance, the official Qwen API service is provided by Qwen Cloud.
In particular, Qwen3.8-27B will be available as a hosted version with more production features, e.g., 1M context length by default, official built-in tools. For more information, please refer to the Qwen3.8-27B Overview. The service is coming soon. Stay tuned for updates.
Following the widespread community adoption of the Qwen3.5 and Qwen3.6 series, we are pleased to introduce Qwen3.8, the most capable generation in the Qwen open-model family to date.
Built on the architectural foundation of Qwen3.5, Qwen3.8 delivers substantial gains across coding, professional work, research, and long-horizon agentic tasks. Qwen3.8-27B brings these advances to a compact, deployment-friendly dense model: a native vision-language model that understands images and videos, with flexible thinking control, designed to carry complex, multi-step tasks through to completion with greater reliability.
Qwen3.8 Highlights
Qwen3.8-27B features the following enhancements:
- Core Capabilities: Comprehensive improvements across coding, professional work, research, and long-horizon agentic tasks.
- Agent Execution: Stronger autonomous planning and better handling of environment feedback, leading to more reliable end-to-end task completion.
- Downstream Compatibility: Broader support for popular harnesses and development tools, making it easier to integrate into your existing stack.
- Flexible Thinking Control: Thinking mode is on by default and can be disabled per request; reasoning depth can be tuned with
reasoning_effort, and reasoning context from historical messages is retained via preserve_thinking.
- Vision-Language Understanding: Native support for image and video understanding, from STEM diagrams and documents to hour-scale videos.
Model Overview
- Type: Causal Language Model with Vision Encoder
- Training Stage: Pre-training & Post-training
- Language Model
- Number of Parameters: 27B
- Hidden Dimension: 5120
- Token Embedding: 248,320 (Padded)
- Number of Layers: 64
- Hidden Layout: 16 × (3 × (Gated DeltaNet → FFN) → 1 × (Gated Attention → FFN))
- Gated DeltaNet:
- Number of Linear Attention Heads: 48 for V and 16 for QK
- Head Dimension: 128
- Gated Attention:
- Number of Attention Heads: 24 for Q and 4 for KV
- Head Dimension: 256
- Rotary Position Embedding Dimension: 64
- Feed Forward Network:
- Intermediate Dimension: 17,408
- LM Output: 248,320 (Padded)
- MTP (Multi-Token Prediction): trained with multiple steps
- Context Length: 262,144 natively and extensible up to 1,000,000 tokens.
Benchmark Results
Text Performance
Quickstart
For streamlined integration, we recommend using Qwen3.8 via APIs.
Serving Qwen3.8
[!Important]
Inference efficiency and throughput vary significantly across frameworks.
We recommend using the latest framework versions to ensure optimal performance and compatibility.
For production workloads or high-throughput scenarios, dedicated serving engines such as SGLang, vLLM, or TokenSpeed are recommended.
Qwen3.8 can be deployed with popular inference frameworks, e.g.:
API Usage
[!Important]
Qwen3.8 models operate in thinking mode by default, generating thinking content signified by <think>\n...</think>\n\n before producing the final response.
To disable thinking content and obtain a direct response, refer to the examples here.
[!Tip]
We recommend using the following sets of sampling parameters for generation:
- Thinking Mode:
temperature=1.0, top_p=0.95, top_k=20, min_p=0.0, presence_penalty=0.0, repetition_penalty=1.0
- Instruct (or non-thinking) mode:
temperature=0.7, top_p=0.80, top_k=20, min_p=0.0, presence_penalty=1.5, repetition_penalty=1.0
Please note that the support for sampling parameters varies according to inference frameworks.
Qwen3.8 comes with official support for reasoning_effort, which can be used to adjust reasoning depth and control cost:
xhigh (default): for complex tasks demanding thorough analysis
medium: balancing accuracy and speed
low: efficient reasoning optimizing for speed and cost
In addition, preserve_thinking is enabled by default for all workloads for the best out-of-the-box experience. To disable preserved thinking, refer to the examples here.
[!Tip]
In multi-turn agentic tasks, lower reasoning effort does not always reduce overall task completion time. Although it may produce faster per-turn responses, it can also lead to insufficient analysis, more failures, and repeated retries, which may increase total latency and token consumption.
Chat Completions API
The Chat Completions API can be used with most inference frameworks, as well as Qwen Cloud.
Before starting, make sure the OpenAI Python SDK is installed and the API key and the API base URL are configured, e.g.:
pip install -U openai # Set the following accordinglyexport OPENAI_BASE_URL='your-base-url'export OPENAI_API_KEY='your-api-key'
Text-Only Input
from openai import OpenAI# Configured by environment variablesclient = OpenAI() messages = [{"role": "user", "content": "Write a Python function to merge two sorted linked lists."}] completion = client.chat.completions.create( model="Qwen/Qwen3.8-27B", messages=messages, extra_body={ "chat_template_kwargs": { "enable_thinking": True, # on by default "preserve_thinking": True, # on by default }, }, reasoning_effort="xhigh", # xhigh by default; supported levels are xhigh, medium, and low stream=True, stream_options={"include_usage": True},) reasoning_content = ""answer_content = ""is_answering = Falseprint("\n" + "=" * 20 + "Reasoning" + "=" * 20 + "\n") for chunk in completion: if not chunk.choices: print("\nUsage:") print(chunk.usage) continue delta = chunk.choices[0].delta if hasattr(delta, "reasoning_content") and delta.reasoning_content is not None: if not is_answering: print(delta.reasoning_content, end="", flush=True) reasoning_content += delta.reasoning_content elif hasattr(delta, "reasoning") and delta.reasoning is not None: if not is_answering: print(delta.reasoning, end="", flush=True) reasoning_content += delta.reasoning if hasattr(delta, "content") and delta.content: if not is_answering: print("\n" + "=" * 20 + "Answer" + "=" * 20 + "\n") is_answering = True print(delta.content, end="", flush=True) answer_content += delta.content messages.append({ "role": "assistant", "content": answer_content, "reasoning_content": reasoning_content, "reasoning": reasoning_content,})
from openai import OpenAI# Configured by environment variablesclient = OpenAI() messages = [ { "role": "user", "content": [ { "type": "image_url", "image_url": { "url": "https://qianwen-res.oss-accelerate.aliyuncs.com/Qwen3.5/demo/CI_Demo/mathv-1327.jpg" } }, { "type": "text", "text": "The centres of the four illustrated circles are in the corners of the square. The two big circles touch each other and also the two little circles. With which factor do you have to multiply the radii of the little circles to obtain the radius of the big circles?\nChoices:\n(A) $\\frac{2}{9}$\n(B) $\\sqrt{5}$\n(C) $0.8 \\cdot \\pi$\n(D) 2.5\n(E) $1+\\sqrt{2}$" } ] }] chat_response = client.chat.completions.create( model="Qwen/Qwen3.8-27B", messages=messages,)print("Chat response:", chat_response)
from openai import OpenAI# Configured by environment variablesclient = OpenAI() messages = [ { "role": "user", "content": [ { "type": "video_url", "video_url": { "url": "https://qianwen-res.oss-accelerate.aliyuncs.com/Qwen3.5/demo/video/N1cdUjctpG8.mp4" } }, { "type": "text", "text": "How many porcelain jars were discovered in the niches located in the primary chamber of the tomb?" } ] }] chat_response = client.chat.completions.create( model="Qwen/Qwen3.8-27B", messages=messages,) # When vLLM is launched with `--media-io-kwargs '{"video": {"num_frames": -1}}'`,# video frame sampling can be configured via `extra_body` (e.g., by setting `fps`).# This feature is currently supported only in vLLM.## By default, `fps=2` and `do_sample_frames=True`.# With `do_sample_frames=True`, you can customize the `fps` value to set your desired video sampling rate.# chat_response = client.chat.completions.create(# model="Qwen/Qwen3.8-27B",# messages=messages,# extra_body={# "mm_processor_kwargs": {"fps": 2, "do_sample_frames": True},# }, # ) print("Chat response:", chat_response)
Instruct (or Non-Thinking) Mode
Qwen3.8-27B will think by default before responding.
You can obtain a direct response from the model without thinking by configuring the API parameters.
For example,
from openai import OpenAI# Configured by environment variablesclient = OpenAI() messages = [ { "role": "user", "content": [ { "type": "image_url", "image_url": { "url": "https://qianwen-res.oss-accelerate.aliyuncs.com/Qwen3.5/demo/RealWorld/RealWorld-04.png" } }, { "type": "text", "text": "Where is this?" } ] }] chat_response = client.chat.completions.create( model="Qwen/Qwen3.8-27B", messages=messages, temperature=0.7, top_p=0.8, presence_penalty=1.5, extra_body={ "top_k": 20, "chat_template_kwargs": {"enable_thinking": False}, }, )print("Chat response:", chat_response)
[!Note]
If you are using APIs from Qwen Cloud, in addition to changing model, please use "enable_thinking": False instead of "chat_template_kwargs": {"enable_thinking": False}.
Disable Preserved Thinking
By default, Qwen3.8 retains thinking blocks from all historical messages, maintaining a complete reasoning trace across the conversation. This behavior, known as preserved thinking, ensures full context continuity and is especially beneficial for agent scenarios where decision consistency and reduced redundant reasoning are critical. It also improves KV cache utilization, optimizing inference efficiency in both thinking and non-thinking modes.
If you prefer to retain only the thinking blocks from the latest user message, you can disable this behavior by setting preserve_thinking to False:
from openai import OpenAI # Configured by environment variablesclient = OpenAI()messages = [...]chat_response = client.chat.completions.create( model="Qwen/Qwen3.8-27B", messages=messages, extra_body={ "chat_template_kwargs": {"preserve_thinking": False}, },)print("Chat response:", chat_response)
[!Note]
If you are using APIs from Qwen Cloud, in addition to changing model, please use "preserve_thinking": False directly instead of wrapping it in chat_template_kwargs.
Best Practices
To achieve optimal performance, we recommend the following settings:
-
Sampling Parameters: We suggest using the following sets of sampling parameters:
- Thinking Mode:
temperature=1.0, top_p=0.95, top_k=20, min_p=0.0, presence_penalty=0.0, repetition_penalty=1.0
- Instruct (or non-thinking) mode:
temperature=0.7, top_p=0.80, top_k=20, min_p=0.0, presence_penalty=1.5, repetition_penalty=1.0
For supported frameworks, you can adjust the presence_penalty parameter between 0 and 2 to reduce endless repetition. However, using a higher value may occasionally result in language mixing and a slight decrease in model performance.
-
Adequate Output Length: To optimize performance on agentic tasks, we recommend allocating sufficient output length to allow the model to generate detailed and comprehensive responses. For frameworks that support separate token limits for internal reasoning and final outputs, we suggest the following configuration within the 1M context length:
- Reasoning Content: Set the maximum output length to 262,144 tokens.
- Final Response: Set the maximum output length to 131,072 tokens.
These settings provide the necessary capacity for complex reasoning while ensuring ample space for high-quality final deliverables.
-
Processing Ultra-Long Texts: Qwen3.8-27B natively supports context lengths of up to 262,144 tokens. For long-horizon tasks where the total length (including both input and output) exceeds this limit, we recommend using RoPE scaling techniques to handle long texts effectively, e.g., YaRN.
YaRN is currently supported by several inference frameworks, e.g., vLLM, SGLang, and TokenSpeed.
In general, there are two approaches to enabling YaRN for supported frameworks:
-
Modifying the model configuration file:
In the config.json file, change the rope_parameters fields in text_config to:
{ "mrope_interleaved": true, "mrope_section": [ 11, 11, 10 ], "rope_type": "yarn", "rope_theta": 10000000, "partial_rotary_factor": 0.25, "factor": 4.0, "original_max_position_embeddings": 262144,}
-
Passing command line arguments:
For vLLM, you can use
VLLM_ALLOW_LONG_MAX_MODEL_LEN=1 vllm serve ... --hf-overrides '{"text_config": {"rope_parameters": {"mrope_interleaved": true, "mrope_section": [11, 11, 10], "rope_type": "yarn", "rope_theta": 10000000, "partial_rotary_factor": 0.25, "factor": 4.0, "original_max_position_embeddings": 262144}}}' --max-model-len 1000000
For SGLang, you can use
SGLANG_ALLOW_OVERWRITE_LONGER_CONTEXT_LEN=1 python -m sglang.launch_server ... --json-model-override-args '{"text_config": {"rope_parameters": {"mrope_interleaved": true, "mrope_section": [11, 11, 10], "rope_type": "yarn", "rope_theta": 10000000, "partial_rotary_factor": 0.25, "factor": 4.0, "original_max_position_embeddings": 262144}}}' --context-length 1000000
For TokenSpeed, you can use
TOKENSPEED_ALLOW_OVERWRITE_LONGER_CONTEXT_LEN=1 tokenspeed serve ... --hf-overrides '{"text_config": {"rope_parameters": {"mrope_interleaved": true, "mrope_section": [11, 11, 10], "rope_type": "yarn", "rope_theta": 10000000, "partial_rotary_factor": 0.25, "factor": 4.0, "original_max_position_embeddings": 262144}}}' --max-model-len 1000000
[!NOTE]
All the notable open-source frameworks implement static YaRN, which means the scaling factor remains constant regardless of input length, potentially impacting performance on shorter texts.
We advise modifying the rope_parameters configuration only when processing long contexts is required.
It is also recommended to modify the factor as needed. For example, if the typical context length for your application is 524,288 tokens, it would be better to set factor as 2.0.
-
Long Video Understanding: To optimize inference efficiency for plain text and images, the size parameter in the released video_preprocessor_config.json is conservatively configured. It is recommended to set the longest_edge parameter in the video_preprocessor_config file to 469,762,048 (corresponding to 224k video tokens) to enable higher frame-rate sampling for hour-scale videos and thereby achieve superior performance. For example,
{"longest_edge": 469762048, "shortest_edge": 4096}
Alternatively, override the default values via engine startup parameters. For implementation details, refer to: vLLM / SGLang.
Citation
If you find our work helpful, feel free to give us a cite.
@misc{qwen38, title = {{Qwen3.8-Max}: A New Bar for Coding and Cowork}, url = {https://qwen.ai/blog?id=qwen3.8}, author = {{Qwen Team}}, month = {August}, year = {2026}}