# Qwen3.8 27B

> Qwen3.8 27B is an open-weight language model from Alibaba (Qwen), released on 14 Aug 2026. It accepts text, images and video and generates text, with a 262K-token context window and up to 33K output tokens per response. As of 4 Oct 2026, the lowest listed API price is $0.99 per million input tokens and $1.49 per million output tokens (Cerebras), across 8 providers we track.

Source page: https://www.aimodel.directory/models/qwen3-8-27b
Last verified: 4 Oct 2026

## Specifications

- **Developer:** Alibaba (Qwen)
- **Release date:** 14 Aug 2026 (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/deepinfra/models/Qwen/Qwen3.8-27B.toml)
- **Status:** Generally available (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/deepinfra/models/Qwen/Qwen3.8-27B.toml)
- **Weights:** Open weights (source: Hugging Face, https://huggingface.co/Qwen/Qwen3.8-27B)
- **License:** apache-2.0 (source: Hugging Face, https://huggingface.co/Qwen/Qwen3.8-27B)
- **Parameters:** 27.8B (source: Hugging Face, https://huggingface.co/Qwen/Qwen3.8-27B)
- **Context window:** 262K (262,144 tokens) (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/deepinfra/models/Qwen/Qwen3.8-27B.toml)
- **Max output:** 33K tokens (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/deepinfra/models/Qwen/Qwen3.8-27B.toml)
- **Input:** Text, images and video (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/deepinfra/models/Qwen/Qwen3.8-27B.toml)
- **Output:** Text (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/deepinfra/models/Qwen/Qwen3.8-27B.toml)
- **Reasoning mode:** Yes (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/deepinfra/models/Qwen/Qwen3.8-27B.toml)
- **Tool calling:** Yes (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/deepinfra/models/Qwen/Qwen3.8-27B.toml)
- **Structured output:** Yes (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/deepinfra/models/Qwen/Qwen3.8-27B.toml)
- **Hugging Face:** Qwen/Qwen3.8-27B (source: Hugging Face, https://huggingface.co/Qwen/Qwen3.8-27B)

## API pricing (USD per 1M tokens)

| Provider | Input | Output | Cached input | Context |
|---|---|---|---|---|
| DeepInfra | $0.20 | $2.50 | $0.05 | 262K |
| Hugging Face Inference Providers | $0.40 | $3 | – | 262K |
| OpenRouter | $0.42 | $2.55 | $0.09 | 1M |
| Cloudflare Workers AI | $0.45 | $3.20 | $0.05 | 262K |
| Nebius AI Studio | $0.45 | $3 | $0.45 | 262K |
| Vercel AI Gateway | $0.50 | $3 | $0.10 | 1M |
| Groq | $0.80 | $4 | – | 131K |
| Cerebras | $0.99 | $1.49 | $0.99 | 131K |

## FAQ

### What is the context window of Qwen3.8 27B?

Qwen3.8 27B has a context window of 262,144 tokens (262K) and can generate up to 32,768 tokens in a single response, according to models.dev.

### How much does the Qwen3.8 27B API cost?

Qwen3.8 27B is listed from $0.99 per million input tokens and $1.49 per million output tokens (Cerebras). Prices last verified 4 Oct 2026.

### Is Qwen3.8 27B open source?

Qwen3.8 27B is an open-weight model: its weights are published on Hugging Face as Qwen/Qwen3.8-27B under the apache-2.0 license. Check the license terms before commercial use.

### When was Qwen3.8 27B released?

Alibaba (Qwen) released Qwen3.8 27B on 14 Aug 2026, according to models.dev.

### Which providers offer Qwen3.8 27B?

We track Qwen3.8 27B on 8 providers: DeepInfra, Hugging Face Inference Providers, OpenRouter, Cloudflare Workers AI, Nebius AI Studio, Vercel AI Gateway, Groq, Cerebras.

---
Data licensed CC BY 4.0. Cite as: AI Model Directory, "Qwen3.8 27B", https://www.aimodel.directory/models/qwen3-8-27b