
Qwen3 235B Instruct
Qwen3 235B Instruct is Alibaba's updated non-thinking Qwen3 flagship, with better instruction following, reasoning, multilingual knowledge and long-context understanding. It's a 235B-parameter mixture of experts with 22B active per token.
Ways to run Qwen3 235B Instruct
Dedicated endpoint
Get a quote →Your own GPUs, tuned for your traffic. We profile your workload, tune the deployment and put throughput and latency in your contract.
Serverless
API docs →Pay per token on shared GPUs, with no capacity to plan or reserve.
Inside your network
Security →Run a dedicated endpoint inside your own network, including air-gapped environments.
Pricing
Use it with the agents we set up for you
Point any open-source agent at your endpoint, or pick from our catalog and we set it up as part of your contract.
Qwen3 235B Instruct FAQ
How much does Qwen3 235B Instruct cost on Saturn Cloud?
Serverless costs $0.075 per 1M input tokens and $0.30 per 1M output tokens. A dedicated endpoint is priced per GPU-hour for the GPUs it runs on, and we size the GPUs to your traffic.
What is Qwen3 235B Instruct's context length?
262,144 tokens natively. With Dual Chunk Attention and MInference it extends to 1,010,000 tokens, which needs config_1m.json and about 1000 GB of GPU memory.
Does Qwen3 235B Instruct think before answering?
No. This 2507 release is non-thinking only and doesn’t produce <think> blocks.
Does Qwen3 235B Instruct support tool calling?
Yes. Qwen recommends Qwen-Agent for tool-call templates and parsers.
What sampling settings does Qwen recommend?
Temperature 0.7, top_p 0.8, top_k 20 and min_p 0. You can set presence_penalty between 0 and 2. Qwen says 16,384 output tokens covers most queries.