Overview
Run Qwen3.8 27B on Crusoe. Fine-tune and serve Alibaba’s newest dense mid-size model with image input, adjustable reasoning effort, and a 262K-token context.
Explore additional models
Frequently
asked questions
Qwen3.8 27B is Alibaba’s newest dense mid-size model, released in August 2026 on the Qwen3.5 architecture. It supports a 262,144-token context, accepts text and image input, and has thinking enabled by default with adjustable effort levels. It ships under the Apache 2.0 license and is available for Serverless Fine-Tuning, Serverless Inference, and Self-Serve Deployments in Crusoe Intelligence Foundry.
No, Qwen3.8 27B is a dense model that uses all 27B parameters on every token. Its mixture-of-experts sibling on Crusoe is Qwen3.6 35B A3B, which activates about 3B parameters per token and is faster and cheaper to run. Choose Qwen3.8 27B when peak quality or image input matters, and Qwen3.6 35B A3B for high-throughput text production.
Qwen3.8 27B delivers strong coding, reasoning, and multilingual performance, and outperforms the mixture-of-experts Qwen3.6 35B A3B on peak quality at the cost of throughput. Its known weakness is long reasoning traces on simple prompts, which the low, medium, and high effort settings let you control.
Yes. Use Serverless Fine-Tuning to customize Qwen3.8 27B on your own data. Upload a JSONL or Parquet dataset, run a LoRA job with pre-configured defaults, then go live in one click using Self-Serve Deployments or download the weights under Apache 2.0. Self-Serve Deployments offer inference profiles optimized for throughput or responsiveness. You can also use Serverless Inference for pay-as-you-go usage, or request a Tailored Deployment for bespoke inference optimization and a dedicated endpoint.
Qwen3.8 27B reasons step by step before answering by default, and a reasoning-effort setting with low, medium, and high levels controls how long that reasoning runs. Lower effort cuts latency and output tokens on simple prompts, while high effort gives the model more room on hard math, code, and multi-step tasks. Set the effort level per request to match the task rather than running every call at maximum.
Yes, Qwen3.8 27B accepts both text and image input and returns text, so it can handle document understanding, chart analysis, screenshot-to-code, and visual question answering alongside standard text tasks. With thinking mode on, it can reason step by step over what it sees, which suits visual tasks that need more than a quick description.


%204.png)
