# Qwen3-Max

> Alibaba's trillion-parameter flagship for coding and agentic tasks

- Maker: Alibaba (Qwen)
- Line: Qwen-Max
- Canonical page: https://ai-tldr.dev/models/qwen3-max/
- Updated: 2026-08-04

## In short

Qwen3-Max is the largest model in Alibaba's Qwen family, crossing the 1-trillion-parameter mark with a Mixture-of-Experts design that activates only a fraction of those parameters per token. Alibaba (Qwen) released it on September 2025 as an API-only model. It has a 262K-token context window. You can use it through Alibaba Cloud Model Studio and OpenRouter.

## Facts

| Fact | Value | Source |
| --- | --- | --- |
| Context window | 262K (max output 32K) | [1] |
| Input price | $1.20 / 1M tokens (0-32K input tier; rises to $2.40 at 32K-128K and $3.00 at 128K-252K) | [2] |
| Output price | $6.00 / 1M tokens (0-32K tier; up to $15.00 at higher tiers) | [2] |
| Licence | Proprietary (closed weights), API access via Alibaba Cloud Model Studio · API only | [1] |
| Released | September 2025 | [3] |
| Parameters | Over 1 trillion (MoE, sparse activation) | [1] |
| Knowledge cutoff | Not officially disclosed | [1] |

## Sources

1. [Alibaba Cloud Model Studio — supported models](https://www.alibabacloud.com/help/en/model-studio/models)
2. [Alibaba Cloud Model Studio — model pricing](https://www.alibabacloud.com/help/en/model-studio/model-pricing)
3. [Dataconomy — Alibaba releases Qwen3-Max](https://dataconomy.com/2025/09/08/alibaba-qwen-3-max-preview/)

## Overview

Qwen3-Max is the largest model in Alibaba's Qwen family, crossing the 1-trillion-parameter mark with a Mixture-of-Experts design that activates only a fraction of those parameters per token. It launched in September 2025 — first as a preview on September 5, then as a full production release later that month — and is positioned as the highest-performing tier of the Qwen-Max line for complex, multi-step tasks. Unlike most of Alibaba's Qwen releases, Qwen3-Max is closed-weight: it runs only through Alibaba Cloud's Model Studio API, Qwen Chat, and third-party gateways such as OpenRouter, with no local deployment.

The model was pretrained on roughly 36 trillion tokens (about twice the data used for Qwen2.5) with a multilingual, coding, and STEM/reasoning emphasis. It ships in two flavors: Qwen3-Max-Instruct, a fast non-thinking model aimed at coding and agentic tool use, and Qwen3-Max-Thinking, a reasoning variant that can call a code interpreter and use parallel test-time compute. Through the API it exposes a 262,144-token context window and up to 32,768 output tokens.

Qwen3-Max is a text-only model — vision tasks in the Qwen3 generation are handled by separate Qwen3-VL models. On release, the preview version of Qwen3-Max reached the global top three on the LMArena text leaderboard, and Alibaba reported strong day-one results on coding and agentic benchmarks. Treat the most aggressive math claims as vendor-reported rather than independently verified.

## Benchmarks

- SWE-bench Verified (Qwen3-Max-Instruct): 69.6% ([source](https://www.marktechpost.com/2025/09/24/alibabas-qwen3-max-production-ready-thinking-mode-1t-parameters-and-day-one-coding-agentic-bench-signals/))
- Tau2-bench (Qwen3-Max-Instruct, agentic tool-calling): 74.8% ([source](https://www.marktechpost.com/2025/09/24/alibabas-qwen3-max-production-ready-thinking-mode-1t-parameters-and-day-one-coding-agentic-bench-signals/))

## Pricing

- Input: $1.20 / 1M tokens (0-32K input tier; rises to $2.40 at 32K-128K and $3.00 at 128K-252K) per 1M tokens
- Output: $6.00 / 1M tokens (0-32K tier; up to $15.00 at higher tiers) per 1M tokens

Tiered by input length on Alibaba Cloud Model Studio (International deployment). Batch invocation is charged at 50% of real-time pricing. Global-deployment pricing is lower.

Source: https://www.alibabacloud.com/help/en/model-studio/model-pricing

## Strengths

- Trillion-parameter MoE scale with efficient sparse activation per token
- Strong software-engineering performance (69.6 on SWE-bench Verified, Instruct variant)
- Top-tier agentic tool-calling (74.8 on Tau2-bench, Instruct variant)
- Large 262K-token context window for long documents and codebases
- Two variants — fast Instruct and reasoning-focused Thinking — under one model line
- Native function calling, structured JSON output, and web-search support via Model Studio

## Best for

- Agentic workflows and multi-tool orchestration
- Software engineering and code generation across large repositories
- Long-context document analysis and summarization
- Complex multi-step reasoning and STEM problem solving
- Multilingual text generation and assistance
- Building API-driven assistants with structured output and function calling

## How to access

- [Alibaba Cloud Model Studio](https://www.alibabacloud.com/help/en/model-studio/models) — model id `qwen3-max`
- [OpenRouter](https://openrouter.ai/qwen/qwen3-max) — model id `qwen/qwen3-max`

## Qwen-Max versions

- [Qwen3.8-Max](https://ai-tldr.dev/models/qwen3-8-max/) (3 August 2026)
- [Qwen3.8-Max-Preview](https://ai-tldr.dev/models/qwen3-8-max-preview/) (19 July 2026)
- [Qwen3.7-Max](https://ai-tldr.dev/models/qwen3-7-max/) (May 2026)
- Qwen3-Max (September 2025) — this page
- [Qwen2.5-Max](https://ai-tldr.dev/models/qwen2-5-max/) (29 January 2025)

## FAQ

### Is Qwen3-Max open source?

No. Unlike many other Qwen models, Qwen3-Max is closed-weight. It is available only through APIs — Alibaba Cloud Model Studio, Qwen Chat, and gateways like OpenRouter — and does not support local deployment.

### How many parameters does Qwen3-Max have?

Alibaba states Qwen3-Max crosses the 1-trillion-parameter mark using a Mixture-of-Experts (MoE) design, so only a sparse subset of those parameters is activated per token. A full technical report with the exact figure was not published at launch.

### What is the context window of Qwen3-Max?

Through Alibaba Cloud Model Studio, Qwen3-Max supports a 262,144-token context window (about 262K) and up to 32,768 output tokens.

### Does Qwen3-Max support images or video?

No. Qwen3-Max is a text-only model. Vision and video tasks in the Qwen3 generation are handled by separate Qwen3-VL models, not by Qwen3-Max.

## Links

- [Qwen3-Max official blog (Qwen)](https://qwen.ai/blog?id=qwen3-max)
- [Alibaba Cloud Model Studio — supported models](https://www.alibabacloud.com/help/en/model-studio/models)
- [Alibaba Cloud Model Studio — model pricing](https://www.alibabacloud.com/help/en/model-studio/model-pricing)
- [OpenRouter — Qwen3-Max](https://openrouter.ai/qwen/qwen3-max)
- [MarkTechPost — Qwen3-Max release analysis & benchmarks](https://www.marktechpost.com/2025/09/24/alibabas-qwen3-max-production-ready-thinking-mode-1t-parameters-and-day-one-coding-agentic-bench-signals/)
- [Dataconomy — Alibaba releases Qwen3-Max](https://dataconomy.com/2025/09/08/alibaba-qwen-3-max-preview/)

Source: AI/TLDR LLM registry, https://ai-tldr.dev/models/qwen3-max/