What is Qwen3.7-Plus?
Qwen3.7-Plus is the mid-tier model of Alibaba's Qwen3.7 series, a 397B mixture-of-experts model with 17B parameters active per token. It reads text, images and video across 1M tokens and is built to act as an agent that reads screens, drives apps and writes code from visual references. Qwen has not released its weights.
Alibaba still recommends it as the balanced choice for coding tools and chat, and it is Qwen Chat's default model. But Alibaba's newer Qwen3.8-Flash and Qwen3.8-27B beat it on the company's own coding tests, and Qwen3.8-Flash costs less.
Is Qwen3.7-Plus still worth using after Qwen3.8?
For free chat and screen-reading agents, yes; for new coding work, test Qwen3.8 first, since Alibaba's own numbers put a cheaper model and an open one ahead of it.
- Still recommended
- Alibaba lists it as the balanced model for coding tools and chat, and Qwen Chat opens on it.
- Qwen3.8-Flash
- Alibaba says its open base model beats Plus on coding and office work after about a ninth of the training cost; the hosted version costs $0.15 / $0.47.
- Qwen3.8-27B
- Scores 61.7 against 57.6 on SWE-bench Pro and 84.3 against 73.3 on OSWorld-Verified, with Apache 2.0 weights.
- Qwen3.8-Max
- Scores higher across Alibaba's vision and agent table, for example 86.1 against 73.3 on OSWorld-Verified, at $2 / $6 per million tokens.
Who is Qwen3.7-Plus best for?
Qwen3.7-Plus is best for agents that read screens and drive apps, free everyday chat in Qwen Chat, budget coding in tools like Claude Code, and long-document work.
Screen and app automation
It scores 79.0 on ScreenSpot Pro and 81.0 on AndroidWorld in Alibaba's table, above Claude Opus 4.6 Max on both.
Free everyday chat
It is the default model in Qwen Chat, which Qwen says is free to use, on the web and in apps for phones and desktops.
Coding tools on a budget
Alibaba recommends it as its balance of cost and quality for coding tools, and it works in Claude Code, OpenClaw and Qwen Code at $0.40 / $1.60 per million tokens.
Long documents
It scores 91.7 on MRCR-v2 at 128K tokens and 91.4 on OmniDocBench 1.5 in Alibaba's tests, with a 1M-token context.
What's included with Qwen3.7-Plus?
A hybrid GUI and CLI agent, image and video input, switchable thinking, built-in web and code tools, OpenAI- and Anthropic-style APIs, and Qwen's Chrome agent.
- Hybrid GUI and CLI agent
- It mixes clicking through interfaces with shell commands and code in one loop; Alibaba's demo ran for 11 hours and wrote over 10,000 lines of an app.
- Image and video input
- It reads screenshots, documents, charts and video, and writes code from a visual reference.
- Thinking on or off
- Thinking can be switched per request, and preserve_thinking keeps reasoning from earlier turns for agent tasks.
- Built-in tools
- Web search, web extraction, code interpreter and image search are available on Qwen Cloud's Responses API.
- OpenAI- and Anthropic-style APIs
- Chat completions and responses APIs, plus an Anthropic-compatible endpoint that Claude Code can use.
- Qwen's Chrome agent
- It powers the Qwen for Chrome extension, which can click, type and navigate in your browser in Agent mode.
How much does Qwen3.7-Plus cost?
Qwen3.7-Plus is free in Qwen Chat and costs $0.40 per million input tokens and $1.60 per million output tokens on Qwen Cloud for prompts up to 256K, or less on Model Studio's Global scope.
- Default model in Qwen Chat
- 1M-token context
- Web, iOS, Android, macOS and Windows apps
- Alibaba Cloud Model Studio, Global deployment scope
- Offered in US (Virginia), Frankfurt, Hong Kong and Tokyo
- Prompts up to 256K; $0.826 / $3.301 up to 1M
- Prompts up to 256K; $1.20 / $4.80 up to 1M
- Cached input $0.08 per million tokens
- 1M free tokens for new accounts, for about 90 days
- Lite: 11,500 credits a month
- Essential $16, Standard $25, Pro $80 a month
- Credits also cover Qwen3.8-Max and other models
Qwen Cloud is Qwen's official API and lists prices in US dollars; Model Studio's US-only scope in US (Virginia) charges the same $0.40 / $1.60. Each request is billed at the tier its input size falls in, and thinking tokens count as output. In China (Beijing) it costs $0.276 / $1.101. Token Plan credits per request depend on the model, tokens, thinking and tools, and plans are not refundable. Full pricing on qwencloud.com. We checked these prices on .
How do readers rate Qwen3.7-Plus?
No one has rated Qwen3.7-Plus yet. Be the first.
Sign in to rate. One rating per account.
Qwen3.7-Plus pros and cons
It is cheap, reads images while it codes and is free in Qwen Chat, but its weights are closed, reviews of its harder coding are mixed, and newer Qwen3.8 models beat it.
Pros
- Users pick it for price: its output costs about half as much as Qwen3.6-Plus, and some say it gets close to far pricier Claude models.
- It reads images while it codes, which users doing visual or design work value over text-only models such as DeepSeek V4 Pro.
- It is free as Qwen Chat's default model, and several users like it for everyday conversation.
- Alibaba's table puts it ahead of GPT-5.4 and Claude Opus 4.6 Max on ScreenSpot Pro, and ahead of Opus and Gemini 3.1 Pro on AndroidWorld.
Cons
- Closed weights: Qwen has not released Qwen3.7-Plus, which many Hacker News and Reddit users criticised given Qwen's history of open models.
- Reviews of harder coding are mixed: some users say it misses issues in code review, needs more detailed instructions than Claude or GPT models, and sometimes uses tools wrongly.
- Alibaba's newer models beat it: Qwen3.8-Flash scores higher on its coding and office tests at $0.15 / $0.47 per million tokens, and the open Qwen3.8-27B does too.
- Some users find it wordy, writing several paragraphs or extra versions where a short answer would do.
What are the best Qwen3.7-Plus alternatives?
Qwen3.8-27B for open weights and higher coding scores, Qwen3.8-Max for Alibaba's strongest model, and Gemini 3.8 Flash for audio input and a free API tier.
Qwen3.8-MaxFreeChoose it when you need Alibaba's strongest model: it scores 86.1 against 73.3 on OSWorld-Verified, costs $2 / $6 per million tokens, and is also free in Qwen Chat.
Gemini 3.8 FlashFreeChoose it when you need audio or PDF input or a free API tier: Google's Flash model reads text, images, video, audio and PDFs across 1M tokens.
Questions about Qwen3.7-Plus
What is Qwen3.7-Plus?
Qwen3.7-Plus is Alibaba's closed, mid-priced multimodal agent model for coding and for driving screens and apps, free as Qwen Chat's default and $0.40 / $1.60 per million tokens on the API.
Who is Qwen3.7-Plus for?
Qwen3.7-Plus is best for agents that read screens and drive apps, free everyday chat in Qwen Chat, budget coding in tools like Claude Code, and long-document work.
Is Qwen3.7-Plus free?
Yes in Qwen Chat, where it is the default model and Qwen says the service is free to use. On the API, new Qwen Cloud accounts get 1 million free tokens for about 90 days; after that it costs $0.40 / $1.60 per million tokens.
How much does Qwen3.7-Plus cost?
Qwen3.7-Plus's paid plans are Model Studio Global at $0.276 / $1.101 per million input / output tokens, Qwen Cloud API at $0.40 / $1.60 per million input / output tokens and Token Plan at From $8 per month.
What are the best Qwen3.7-Plus alternatives?
The closest alternatives on Recaply are Qwen3.8-Max and Gemini 3.8 Flash.
Where can I try Qwen3.7-Plus?
On its official website, qwen.ai.
Is Qwen3.7-Plus open source?
No. Qwen has not released Qwen3.7-Plus weights, so you can use it only through Qwen Chat, Qwen Cloud or Alibaba Cloud Model Studio. Alibaba's newer open models in this range are Qwen3.8-27B and Qwen3.8-Flash-Next.
How big is Qwen3.7-Plus?
Alibaba lists it as a 397B-parameter mixture-of-experts model with 17B parameters active per token, in its comparison table for Qwen3.8-Flash-Next. That makes it far larger than Qwen3.8-27B, though the 27B model scores higher on Alibaba's coding tests.
Can I use Qwen3.7-Plus in Claude Code?
Yes. Alibaba's API accepts the Anthropic format, so you point Claude Code at the Anthropic-compatible endpoint, set the model to qwen3.7-plus and use your API key. OpenClaw and Qwen Code work as well, and Qwen's Token Plan covers these tools.
Should I use Qwen3.8-Flash instead of Qwen3.7-Plus?
For coding and office agents, Alibaba's numbers point that way: Qwen3.8-Flash-Next, the open model behind Qwen3.8-Flash, beats Plus on CoWorkBench (73.9 against 65.1) and Toolathlon Verified (73.5 against 50.6), and Qwen3.8-Flash costs $0.15 / $0.47 per million tokens. Test both on your own tasks.
Our verdict
Qwen3.7-Plus is a sensible low-cost agent model, but new projects should test Qwen3.8 first.
It reads screens well and costs little, yet Alibaba's own tables show Qwen3.8-Flash and the open Qwen3.8-27B beating it on coding, with Flash cheaper still.
- Skip it for
- Anyone who needs weights they can host or audio input, and teams that want the strongest coding model.
- Check first
- Run your own tasks against Qwen3.8-Flash before you commit to Plus pricing.
Qwen3.6-Plus
Qwen3.8-Flash
Qwen3.6-Max-Preview
Qwen3.7-Max
Qwen3.5 Small
Qwen3-Max-Thinking