Marketplace · Text & chat
Qwen3-30B-A3B-Instruct-2507-Q4_K_M
Call Qwen3-30B-A3B-Instruct-2507-Q4_K_M through one OpenAI-compatible API. It's served by 1 host on the Inpherio grid, each at a price it sets — your request goes to the cheapest healthy one, and fails over to the next if it's down.
At a glance.
from openai import OpenAI
client = OpenAI(base_url="https://inpherio.co.uk/v1",
api_key=INPHERIO_API_KEY)
reply = client.chat.completions.create(
model="Qwen3-30B-A3B-Instruct-2507-Q4_K_M",
messages=[{"role": "user", "content": "Hello"}],
)Qwen3-30B-A3B-Instruct-2507-Q4_K_M in detail.
We introduce the updated version of the **Qwen3-30B-A3B non-thinking mode**, named **Qwen3-30B-A3B-Instruct-2507**, featuring the following key enhancements:
- Published by
- Qwen
- Parameters
- 30.5B
- Format
- transformers
- Architecture
- qwen3_moe
- Licence
- Apache 2.0
- Task
- text generation
- Library
- transformers
- First published
- 28 Jul 2025
- Last updated
- 17 Sep 2025
- Downloads (30 days)
- 820,521 on Hugging Face
- Likes
- 856
qwen3_moe text-generation
The model card
Published by the model's authors on Hugging Face — shown as they wrote it, unedited by Inpherio.
Qwen3-30B-A3B-Instruct-2507
Highlights
We introduce the updated version of the Qwen3-30B-A3B non-thinking mode, named Qwen3-30B-A3B-Instruct-2507, featuring the following key enhancements:
- Significant improvements in general capabilities, including instruction following, logical reasoning, text comprehension, mathematics, science, coding and tool usage.
- Substantial gains in long-tail knowledge coverage across multiple languages.
- Markedly better alignment with user preferences in subjective and open-ended tasks, enabling more helpful responses and higher-quality text generation.
- Enhanced capabilities in 256K long-context understanding.
Model Overview
Qwen3-30B-A3B-Instruct-2507 has the following features:
- Type: Causal Language Models
- Training Stage: Pretraining & Post-training
- Number of Parameters: 30.5B in total and 3.3B activated
- Number of Paramaters (Non-Embedding): 29.9B
- Number of Layers: 48
- Number of Attention Heads (GQA): 32 for Q and 4 for KV
- Number of Experts: 128
- Number of Activated Experts: 8
- Context Length: 262,144 natively.
NOTE: This model supports only non-thinking mode and does not generate ``` blocks in its output. Meanwhile, specifying enable_thinking=False` is no longer required.
For more details, including benchmark evaluation, hardware requirements, and inference performance, please refer to our blog, GitHub, and Documentation.
Performance
| Deepseek-V3-0324 | GPT-4o-0327 | Gemini-2.5-Flash Non-Thinking | Qwen3-235B-A22B Non-Thinking | Qwen3-30B-A3B Non-Thinking | Qwen3-30B-A3B-Instruct-2507 | |
|---|---|---|---|---|---|---|
| Knowledge | ||||||
| MMLU-Pro | 81.2 | 79.8 | 81.1 | 75.2 | 69.1 | 78.4 |
| MMLU-Redux | 90.4 | 91.3 | 90.6 | 89.2 | 84.1 | 89.3 |
| GPQA | 68.4 | 66.9 | 78.3 | 62.9 | 54.8 | 70.4 |
| SuperGPQA | 57.3 | 51.0 | 54.6 | 48.2 | 42.2 | 53.4 |
| Reasoning | ||||||
| AIME25 | 46.6 | 26.7 | 61.6 | 24.7 | 21.6 | 61.3 |
| HMMT25 | 27.5 | 7.9 | 45.8 | 10.0 | 12.0 | 43.0 |
| ZebraLogic | 83.4 | 52.6 | 57.9 | 37.7 | 33.2 | 90.0 |
| LiveBench 20241125 | 66.9 | 63.7 | 69.1 | 62.5 | 59.4 | 69.0 |
| Coding | ||||||
| LiveCodeBench v6 (25.02-25.05) | 45.2 | 35.8 | 40.1 | 32.9 | 29.0 | 43.2 |
| MultiPL-E | 82.2 | 82.7 | 77.7 | 79.3 | 74.6 | 83.8 |
| Aider-Polyglot | 55.1 | 45.3 | 44.0 | 59.6 | 24.4 | 35.6 |
| Alignment | ||||||
| IFEval | 82.3 | 83.9 | 84.3 | 83.2 | 83.7 | 84.7 |
| Arena-Hard v2* | 45.6 | 61.9 | 58.3 | 52.0 | 24.8 | 69.0 |
| Creative Writing v3 | 81.6 | 84.9 | 84.6 | 80.4 | 68.1 | 86.0 |
| WritingBench | 74.5 | 75.5 | 80.5 | 77.0 | 72.2 | 85.5 |
| Agent | ||||||
| BFCL-v3 | 64.7 | 66.5 | 66.1 | 68.0 | 58.6 | 65.1 |
| TAU1-Retail | 49.6 | 60.3# | 65.2 | 65.2 | 38.3 | 59.1 |
| TAU1-Airline | 32.0 | 42.8# | 48.0 | 32.0 | 18.0 | 40.0 |
| TAU2-Retail | 71.1 | 66.7# | 64.3 | 64.9 | 31.6 | 57.0 |
| TAU2-Airline | 36.0 | 42.0# | 42.5 | 36.0 | 18.0 | 38.0 |
| TAU2-Telecom | 34.0 | 29.8# | 16.9 | 24.6 | 18.4 | 12.3 |
| Multilingualism | ||||||
| MultiIF | 66.5 | 70.4 | 69.4 | 70.2 | 70.8 | 67.9 |
| MMLU-ProX | 75.8 | 76.2 | 78.3 | 73.2 | 65.1 | 72.0 |
| INCLUDE | 80.1 | 82.1 | 83.8 | 75.6 | 67.8 | 71.9 |
| PolyMATH | 32.2 | 25.5 | 41.9 | 27.0 | 23.3 | 43.1 |
*: For reproducibility, we report the win rates evaluated by GPT-4.1.
\#: Results were generated using GPT-4o-20241120, as access to the native function calling API of GPT-4o-0327 was unavailable.
Read the rest of the model card
Quickstart
The code of Qwen3-MoE has been in the latest Hugging Face transformers and we advise you to use the latest version of transformers.
With transformers<4.51.0, you will encounter the following error:
KeyError: 'qwen3_moe'
The following contains a code snippet illustrating how to use the model generate content based on given inputs.
from transformers import AutoModelForCausalLM, AutoTokenizer
model_name = "Qwen/Qwen3-30B-A3B-Instruct-2507"
### load the tokenizer and the model
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
model_name,
torch_dtype="auto",
device_map="auto"
)
### prepare the model input
prompt = "Give me a short introduction to large language model."
messages = [
{"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True,
)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
### conduct text completion
generated_ids = model.generate(
**model_inputs,
max_new_tokens=16384
)
output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist()
content = tokenizer.decode(output_ids, skip_special_tokens=True)
print("content:", content)
For deployment, you can use sglang>=0.4.6.post1 or vllm>=0.8.5 or to create an OpenAI-compatible API endpoint:
- SGLang: ``
shell python -m sglang.launch_server --model-path Qwen/Qwen3-30B-A3B-Instruct-2507 --context-length 262144`` - vLLM: ``
shell vllm serve Qwen/Qwen3-30B-A3B-Instruct-2507 --max-model-len 262144``
Note: If you encounter out-of-memory (OOM) issues, consider reducing the context length to a shorter value, such as 32,768.
For local use, applications such as Ollama, LMStudio, MLX-LM, llama.cpp, and KTransformers have also supported Qwen3.
Agentic Use
Qwen3 excels in tool calling capabilities. We recommend using Qwen-Agent to make the best use of agentic ability of Qwen3. Qwen-Agent encapsulates tool-calling templates and tool-calling parsers internally, greatly reducing coding complexity.
To define the available tools, you can use the MCP configuration file, use the integrated tool of Qwen-Agent, or integrate other tools by yourself.
from qwen_agent.agents import Assistant
### Define LLM
llm_cfg = {
'model': 'Qwen3-30B-A3B-Instruct-2507',
# Use a custom endpoint compatible with OpenAI API:
'model_server': 'http://localhost:8000/v1', # api_base
'api_key': 'EMPTY',
}
### Define Tools
tools = [
{'mcpServers': { # You can specify the MCP configuration file
'time': {
'command': 'uvx',
'args': ['mcp-server-time', '--local-timezone=Asia/Shanghai']
},
"fetch": {
"command": "uvx",
"args": ["mcp-server-fetch"]
}
}
},
'code_interpreter', # Built-in tools
]
### Define Agent
bot = Assistant(llm=llm_cfg, function_list=tools)
### Streaming generation
messages = [{'role': 'user', 'content': 'https://qwenlm.github.io/blog/ Introduce the latest developments of Qwen'}]
for responses in bot.run(messages=messages):
pass
print(responses)
Processing Ultra-Long Texts
To support ultra-long context processing (up to 1 million tokens), we integrate two key techniques:
- **Dual Chunk Attention (DCA)**: A length extrapolation method that splits long sequences into manageable chunks while preserving global coherence.
- **MInference**: A sparse attention mechanism that reduces computational overhead by focusing on critical token interactions.
Together, these innovations significantly improve both generation quality and inference efficiency for sequences beyond 256K tokens. On sequences approaching 1M tokens, the system achieves up to a 3× speedup compared to standard attention implementations.
For full technical details, see the Qwen2.5-1M Technical Report.
How to Enable 1M Token Context
[!NOTE] To effectively process a 1 million token context, users will require approximately 240 GB of total GPU memory. This accounts for model weights, KV-cache storage, and peak activation memory demands.
Step 1: Update Configuration File
Download the model and replace the content of your config.json with config_1m.json, which includes the config for length extrapolation and sparse attention.
export MODELNAME=Qwen3-30B-A3B-Instruct-2507
huggingface-cli download Qwen/${MODELNAME} --local-dir ${MODELNAME}
mv ${MODELNAME}/config.json ${MODELNAME}/config.json.bak
mv ${MODELNAME}/config_1m.json ${MODELNAME}/config.json
Step 2: Launch Model Server
After updating the config, proceed with either vLLM or SGLang for serving the model.
Option 1: Using vLLM
To run Qwen with 1M context support:
pip install -U vllm \
--torch-backend=auto \
--extra-index-url https://wheels.vllm.ai/nightly
Then launch the server with Dual Chunk Flash Attention enabled:
VLLM_ATTENTION_BACKEND=DUAL_CHUNK_FLASH_ATTN VLLM_USE_V1=0 \
vllm serve ./Qwen3-30B-A3B-Instruct-2507 \
--tensor-parallel-size 4 \
--max-model-len 1010000 \
--enable-chunked-prefill \
--max-num-batched-tokens 131072 \
--enforce-eager \
--max-num-seqs 1 \
--gpu-memory-utilization 0.85
Key Parameters
| Parameter | Purpose |
|---|---|
VLLM_ATTENTION_BACKEND=DUAL_CHUNK_FLASH_ATTN | Enables the custom attention kernel for long-context efficiency |
--max-model-len 1010000 | Sets maximum context length to ~1M tokens |
--enable-chunked-prefill | Allows chunked prefill for very long inputs (avoids OOM) |
--max-num-batched-tokens 131072 | Controls batch size during prefill; balances throughput and memory |
--enforce-eager | Disables CUDA graph capture (required for dual chunk attention) |
--max-num-seqs 1 | Limits concurrent sequences due to extreme memory usage |
--gpu-memory-utilization 0.85 | Set the fraction of GPU memory to be used for the model executor |
Option 2: Using SGLang
First, clone and install the specialized branch:
git clone https://github.com/sgl-project/sglang.git
cd sglang
pip install -e "python[all]"
Launch the server with DCA support:
python3 -m sglang.launch_server \
--model-path ./Qwen3-30B-A3B-Instruct-2507 \
--context-length 1010000 \
--mem-frac 0.75 \
--attention-backend dual_chunk_flash_attn \
--tp 4 \
--chunked-prefill-size 131072
Key Parameters
| Parameter | Purpose |
|---|---|
--attention-backend dual_chunk_flash_attn | Activates Dual Chunk Flash Attention |
--context-length 1010000 | Defines max input length |
--mem-frac 0.75 | The fraction of the memory used for static allocation (model weights and KV cache memory pool). Use a smaller value if you see out-of-memory errors. |
--tp 4 | Tensor parallelism size (matches model sharding) |
--chunked-prefill-size 131072 | Prefill chunk size for handling long inputs without OOM |
Troubleshooting:
- Encountering the error: "The model's max sequence length (xxxxx) is larger than the maximum number of tokens that can be stored in the KV cache." or "RuntimeError: Not enough memory. Please try to increase --mem-fraction-static."
The VRAM reserved for the KV cache is insufficient.
- vLLM: Consider reducing the `
max_model_lenor increasing thetensor_parallel_sizeandgpu_memory_utilization. Alternatively, you can reducemax_num_batched_tokens`, although this may significantly slow down inference. - SGLang: Consider reducing the `
context-lengthor increasing thetpandmem-frac. Alternatively, you can reducechunked-prefill-size`, although this may significantly slow down inference.
- Encountering the error: "torch.OutOfMemoryError: CUDA out of memory."
The VRAM reserved for activation weights is insufficient. You can try lowering `gpu_memory_utilization or mem-frac`, but be aware that this might reduce the VRAM available for the KV cache.
- Encountering the error: "Input prompt (xxxxx tokens) + lookahead slots (0) is too long and exceeds the capacity of the block manager." or "The input (xxx xtokens) is longer than the model's context length (xxx tokens)."
The input is too lengthy. Consider using a shorter sequence or increasing the `max_model_len or context-length`.
Long-Context Performance
We test the model on an 1M version of the RULER benchmark.
| Model Name | Acc avg | 4k | 8k | 16k | 32k | 64k | 96k | 128k | 192k | 256k | 384k | 512k | 640k | 768k | 896k | 1000k |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Qwen3-30B-A3B (Non-Thinking) | 72.0 | 97.1 | 96.1 | 95.0 | 92.2 | 82.6 | 79.7 | 76.9 | 70.2 | 66.3 | 61.9 | 55.4 | 52.6 | 51.5 | 52.0 | 50.9 |
| Qwen3-30B-A3B-Instruct-2507 (Full Attention) | 86.8 | 98.0 | 96.7 | 96.9 | 97.2 | 93.4 | 91.0 | 89.1 | 89.8 | 82.5 | 83.6 | 78.4 | 79.7 | 77.6 | 75.7 | 72.8 |
| Qwen3-30B-A3B-Instruct-2507 (Sparse Attention) | 86.8 | 98.0 | 97.1 | 96.3 | 95.1 | 93.6 | 92.5 | 88.1 | 87.7 | 82.9 | 85.7 | 80.7 | 80.0 | 76.9 | 75.5 | 72.2 |
- All models are evaluated with Dual Chunk Attention enabled.
- Since the evaluation is time-consuming, we use 260 samples for each length (13 sub-tasks, 20 samples for each).
Best Practices
To achieve optimal performance, we recommend the following settings:
- Sampling Parameters:
- We suggest using
Temperature=0.7,TopP=0.8,TopK=20, andMinP=0. - For supported frameworks, you can adjust the
presence_penaltyparameter between 0 and 2 to reduce endless repetitions. However, using a higher value may occasionally result in language mixing and a slight decrease in model performance.
- Adequate Output Length: We recommend using an output length of 16,384 tokens for most queries, which is adequate for instruct models.
- Standardize Output Format: We recommend using prompts to standardize model outputs when benchmarking.
- Math Problems: Include "Please reason step by step, and put your final answer within \boxed{}." in the prompt.
- Multiple-Choice Questions: Add the following JSON structure to the prompt to standardize responses: "Please show your choice in the
answerfield with only the choice letter, e.g.,"answer": "C"."
Citation
If you find our work helpful, feel free to give us a cite.
@misc{qwen3technicalreport,
title={Qwen3 Technical Report},
author={Qwen Team},
year={2025},
eprint={2505.09388},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2505.09388},
}Sources: Qwen/Qwen3-30B-A3B-Instruct-2507 on Hugging Face. Collected from public sources on 14 Nov 2026; the host serving it may run a different build or quantisation — the marketplace shows each host's own measurements.
Who's serving it.
Every host benchmarked this model on its own hardware before serving it. Prices are set by each host; health is scored by the platform from what it observes.
| Host | Tier | Price | Measured speed | Health | State | Rent |
|---|---|---|---|---|---|---|
| iPhone 14 Pro Max · on-device | L1 | £0.03 in · £0.06 out | 2 tok/s · 498 ms first token | 79 | loads on request | £0.10/h |
Live data, refreshed on every visit. How billing works.
Room on the grid for your machine.
Right now one machine serves Qwen3-30B-A3B-Instruct-2507-Q4_K_M. If it drops, so does every request for it — a second host is what turns it into failover, and the grid needs that more than another copy. The router weighs price, measured speed, health and whether the model is already loaded — so a host with a sharper price or quicker hardware wins the next call.
Set your own price — you keep 85% of what it earns, and new hosts start with 12 months fee-free (up to £250 of earnings). A payout fee of £2 + 1% per payout to a UK bank account, £2 + 2% to any other country applies when you're paid. Install the connector, load Qwen3-30B-A3B-Instruct-2507-Q4_K_M, and the platform benchmarks it on your hardware before your first request.