---
last_updated: 2026-11-21
source_url: https://opper.ai/alibaba/qwen3-8-flash
title: "Qwen3.8 Flash by Alibaba"
description: "Qwen3.8-Flash is Alibaba's fast multimodal Qwen model with a 1M-token context, built on the Qwen4-preview Qwen3.8-Flash-Next architecture for coding, agentic workflows and image and video understanding."
---

# Qwen3.8 Flash

By [Alibaba](https://opper.ai/alibaba).

Released 26 Aug 2026 · [Announcement](https://docs.qwencloud.com/changelog/models) · [All Alibaba releases](https://opper.ai/model-releases/alibaba)

Qwen3.8-Flash is the Qwen team's production model based on the open-weight Qwen3.8-Flash-Next, which Alibaba released in August 2026 as an experimental preview of the architecture that will underpin Qwen4. Qwen describes it as the latest multimodal model in the family, combining strong reasoning and generation with speed. It accepts text, images and video, outputs text, and supports a context window of roughly 1 million tokens with up to 131,072 output tokens.

The underlying Flash-Next design is a Mixture-of-Experts model with 125B total and 6B activated parameters, plus a 51B n-gram embedding. It combines hybrid attention (Gated DeltaNet with Qwen Sparse Attention), gated residuals and 512 experts across 48 layers. The model thinks by default, with thinking switchable and adjustable through reasoning effort, and supports function calling and structured outputs. On Qwen's numbers, Flash-Next scores 91.7 on GPQA Diamond, 62.5 on SWE-bench Pro, 73.9 on CoWorkBench and 84.5 on AndroidWorld.

Alibaba highlights coding assistance, agentic workflows and visual understanding, such as fixing code autonomously, operating desktop applications and analyzing charts and long videos. It fits long-document and codebase work where a large context and low active-parameter count matter.

## Available routes

Qwen3.8 Flash runs on 3 different routes through the Opper gateway. Prices are per 1M tokens. In the EU it is served by Alibaba Cloud (Germany). [See every Qwen model hosted in the EU](https://opper.ai/models/eu/qwen). Its routes and prices, live from the catalogue, are public JSON with no API key needed at https://api.opper.ai/v3/models?q=qwen3.8-flash.

| Provider | Model id | Region | Zero data retention | Training | Input | Output | Cache read |
| --- | --- | --- | --- | --- | --- | --- | --- |
| [Alibaba Cloud](https://opper.ai/provider/alibaba) | `alibaba:eu/qwen3.8-flash` | EU (Germany) | Abuse monitoring | No | $0.16 | $0.51 | Input price |
| [DeepInfra](https://opper.ai/provider/deepinfra) | `deepinfra/Qwen/Qwen3.8-Flash` | US (United States) | Yes | No | $0.11 | $0.38 | $0.01 |
| [Novita](https://opper.ai/provider/novita) | `novita/qwen/qwen3.8-flash` | US (United States) | Yes | No | $0.15 | $0.47 | $0.02 |

## Key info

- Model id: `qwen3.8-flash` (routes across Alibaba Cloud, Novita)
- Input: Text, Image, Video
- Output: Text
- Features: Reasoning, Tools, Structured output
- Context window: 1M
- Max output: 131K
- Input price: $0.16 per 1M tokens
- Output price: $0.51 per 1M tokens
- Released: 26 Aug 2026
- EU residency available on Alibaba Cloud
- US residency available on DeepInfra and Novita
- Zero data retention on pay-as-you-go
- No training by default

## Data handling per route

Each route hosting Qwen3.8 Flash has its own privacy posture, residency, and GDPR terms. Postures are maintained by Opper with a last-verification timestamp.

### Alibaba Cloud — Germany

Zero data retention with abuse monitoring: prompts and outputs are kept for abuse monitoring only, for a window the provider does not state, and are not used for training. No training on customer data. EU.

- Zero data retention: With abuse monitoring, window not stated.
- Training: No training on customer data.
- Logging: Abuse monitoring
- Abuse monitoring: On by default, holds flagged content
- Caching: Content cached for replay
- Transfer mechanism: Not applicable — data stays in EU

Verified by Opper on 2026-11-16.

### DeepInfra — United States

Zero data retention: nothing is logged, held for abuse monitoring or used for training. No training on customer data. US; SCCs.

- Zero data retention: Yes. Nothing is logged, held for abuse monitoring or used for training.
- Training: No training on customer data.
- Logging: None
- Abuse monitoring: No classifier
- Transfer mechanism: SCCs

Verified by Opper on 2026-10-19.

### Novita — United States

Zero data retention: nothing is logged, held for abuse monitoring or used for training. No training on customer data. US; SCCs.

- Zero data retention: Yes. Nothing is logged, held for abuse monitoring or used for training.
- Training: No training on customer data.
- Logging: None
- Abuse monitoring: On by default, keeps nothing
- Transfer mechanism: SCCs

Verified by Opper on 2026-09-15.

## Benchmarks

Independent benchmark scores via [Artificial Analysis](https://artificialanalysis.ai) ([view on AA](https://artificialanalysis.ai/models/qwen3-8-flash-next)). Indices and domain scores are out of 100, Omniscience runs from −100 to 100, Briefcase is an Elo rating, and the other evaluation scores are the share of tasks solved.

- Global rank: #35 of 520 LLMs on the [LLM leaderboard](https://opper.ai/llm-leaderboard)
- Tier: Efficient
- Output speed: 58 tok/s
- Intelligence Index: 28.6
- Coding: 26.4
- Agentic: 40.6

### Coding

- FrontierCode 2.0: 27%
- Terminal-Bench 4.0: 25%
- SciCode: 51%

### Agentic & knowledge work

- AutomationBench: 56%
- GDPval: 57%
- GDP.pdf: 16%
- Briefcase (Elo): 1568

### Science & reasoning

- Terminal-Bench Science: 7%
- CritPt (hard): 25%
- HLE Diamond: 16%
- Humanity's Last Exam: 38%
- GPQA Diamond: 92%
- Long-context reasoning: 80%

### Knowledge

- Omniscience: -12.2
- MMMU-Pro: 79%

### By domain

- Finance & accounting: 39.0
- Legal: 35.1
- Healthcare & medical: 36.4
- Engineering: 33.0
- Economics: 36.1
- Strategy & operations: 44.5

## Get started

Call Qwen3.8 Flash through the Opper gateway with one API key. Opper is drop-in compatible with the OpenAI, Anthropic and Google AI SDKs.

Qwen3.8 Flash is a premium model on Opper. Sign up needs no credit card: you get an API key straight away and the free models work in the playground and the API. Add a card to use premium models, pay-as-you-go with no minimum.

### Set it up with your agent

Paste this into a coding agent like Claude Code, Cursor or Codex and it will wire up Opper for you:

```text
Use curl to download, read and follow: https://skills.opper.ai Call Qwen3.8 Flash
```

### Or call it directly

OpenAI:

```javascript
import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.OPPER_API_KEY,
  baseURL: "https://api.opper.ai/v3/compat",
});

const completion = await client.chat.completions.create({
  model: "qwen3.8-flash",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(completion.choices[0].message.content);
```

Anthropic:

```javascript
import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: process.env.OPPER_API_KEY,
  baseURL: "https://api.opper.ai/v3/compat",
});

const message = await client.messages.create({
  model: "qwen3.8-flash",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello" }],
});
console.log(message.content);
```

Gemini:

```javascript
import { GoogleGenAI } from "@google/genai";

const ai = new GoogleGenAI({
  apiKey: process.env.OPPER_API_KEY,
  httpOptions: { baseUrl: "https://api.opper.ai/v3/compat" },
});

const response = await ai.models.generateContent({
  model: "qwen3.8-flash",
  contents: "Hello",
});
console.log(response.text);
```

cURL:

```bash
curl https://api.opper.ai/v3/compat/chat/completions \
  -H "Authorization: Bearer $OPPER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.8-flash",
    "messages": [{ "role": "user", "content": "Hello" }]
  }'
```

## Use Qwen3.8 Flash in your apps

Pick your tool to see how to run it on Qwen3.8 Flash through Opper.

- [Claude Code](https://opper.ai/apps/claude-code/qwen3-8-flash)
- [Claude Cowork](https://opper.ai/apps/claude-cowork/qwen3-8-flash)
- [Codex](https://opper.ai/apps/codex/qwen3-8-flash)
- [DeepSeek Harness](https://opper.ai/apps/deepseek-harness/qwen3-8-flash)
- [GitHub Copilot CLI](https://opper.ai/apps/github-copilot-cli/qwen3-8-flash)
- [Aider](https://opper.ai/apps/aider/qwen3-8-flash)
- [Goose](https://opper.ai/apps/goose/qwen3-8-flash)
- [Droid](https://opper.ai/apps/droid/qwen3-8-flash)
- [OpenCode](https://opper.ai/apps/opencode/qwen3-8-flash)
- [Pi](https://opper.ai/apps/pi/qwen3-8-flash)
- [OpenClaw](https://opper.ai/apps/openclaw/qwen3-8-flash)
- [Hermes](https://opper.ai/apps/hermes/qwen3-8-flash)
- [Copilot for Obsidian](https://opper.ai/apps/obsidian-copilot/qwen3-8-flash)
- [OmniRoute](https://opper.ai/apps/omniroute/qwen3-8-flash)
- [Cursor](https://opper.ai/apps/cursor/qwen3-8-flash)
- [Cline](https://opper.ai/apps/cline/qwen3-8-flash)
- [Continue](https://opper.ai/apps/continue/qwen3-8-flash)
- [Kilo Code](https://opper.ai/apps/kilo-code/qwen3-8-flash)
- [GitHub Copilot](https://opper.ai/apps/github-copilot/qwen3-8-flash)
- [Zed](https://opper.ai/apps/zed/qwen3-8-flash)
- [JetBrains AI Assistant](https://opper.ai/apps/jetbrains-ai-assistant/qwen3-8-flash)
- [Voidleap Code](https://opper.ai/apps/voidleap-code/qwen3-8-flash)
- [n8n](https://opper.ai/apps/n8n/qwen3-8-flash)
- [LibreChat](https://opper.ai/apps/librechat/qwen3-8-flash)
- [Open WebUI](https://opper.ai/apps/open-webui/qwen3-8-flash)
- [See all apps](https://opper.ai/apps)

## Other models from Alibaba

- [Qwen4-Max](https://opper.ai/alibaba/qwen4-max)
- [Qwen3.8 Max](https://opper.ai/alibaba/qwen3-8-max)
- [Qwen3.8 27B](https://opper.ai/alibaba/qwen3-8-27b)
- [Qwen3.8-2.4T-A95B](https://opper.ai/alibaba/qwen3-8-2-4t-a95b)
- [Qwen3.8 Flash-Next](https://opper.ai/alibaba/qwen3-8-flash-next)
- [Qwen3.7-Max](https://opper.ai/alibaba/qwen3-7-max)
