Architecture: what makes it capable and efficient
Qwen3.7 Plus uses a sparse Mixture-of-Experts design. Rather than activating the full parameter space on every token, the model routes each token to a small set of specialized expert sub-networks. This keeps inference compute proportional to the active parameter count, not the total model size.
Sparse MoE (mixture of experts)Each token is routed to a specialized subset of the model's expert networks. Only a fraction of total parameters are active per forward pass, which keeps latency and cost well below what a comparable dense model would require.
Native multimodal fusionText, images, and video are processed in a single forward pass. There are no separate vision adapters or preprocessing pipelines — the model was trained from scratch on multimodal tokens, allowing natural cross-modal reasoning without stitching together separate models.
Configurable reasoning depthCallers can dial reasoning intensity up or down per request. At low settings the model behaves as a fast instruction-follower; at higher settings it performs multi-step chain-of-thought decomposition suited for coding, visual analysis, and agentic planning tasks.
Full tool use supportFunction calling is a production default, not an experimental add-on. Single tool calls, auto tool choice, JSON mode, and vision + tools are all supported out of the box.
Core capabilities
Text, image & video inputInclude screenshots, charts, short video clips, and documents alongside text in the same prompt. Inputs are natively understood — not converted or approximated by a separate model.
Tool use and function callingCall external APIs, retrieve structured data, and chain multi-step operations. OpenAI-compatible function calling format means existing tool integrations work without modification.
Thinking preservationReasoning traces persist across conversation turns, reducing redundant computation in iterative development workflows and multi-step planning tasks.
Coding and agentic workflowsThe model understands project-level context, generates well-structured code across languages, and participates in autonomous agent loops — planning, calling tools, and iterating without human intervention between steps.
201-language supportStrong multilingual coverage with consistent instruction-following quality across languages including English, Arabic, Japanese, Chinese, and Estonian.
Benchmark performance
Who should use Qwen3.7 Plus?
Teams building visual AI productsApplications that need to process images, screenshots, charts, or video alongside text — from document intelligence to UI-aware agents — without managing a separate vision model.
Agent developersEngineers building multi-step tool-use agents that require consistent structured output, long state histories, reliable multi-turn reasoning, and visual context awareness — without the cost of a tier-1 model.
Coding infrastructure teamsTeams running code generation, review, or refactoring pipelines that benefit from reasoning depth and multimodal inputs, such as processing screenshots of error states or UI mockups.
Production-scale appsApplications running millions of requests per month where per-token cost matters. At $0.52/M input, Qwen3.7 Plus delivers reasoning and vision capabilities at a fraction of frontier model pricing.
Research and analysis pipelinesTeams that combine text and visual inputs for competitive analysis, document review, and multi-source synthesis — where both modalities need to be processed in a single coherent pass.
Measured on the AI/ML API gateway across live traffic. Figures are medians unless stated. Last updated 2026-10-12.
Read and explain screenshots or documents, including visual materials such as charts and tables.
Support workflow automation and tool calling.
Use web search to answer time-sensitive questions.
Help with visual-to-code or UI-related tasks.
import requests r = requests.post( "https://api.aimlapi.com/v1/chat/completions", headers={"Authorization": "Bearer " + AIMLAPI_KEY}, json={ "model": "alibaba/qwen3.7-plus", "messages": [ { "role": "user", "content": "Hello!" } ] }, ) print(r.json())
const r = await fetch("https://api.aimlapi.com/v1/chat/completions", { method: "POST", headers: { Authorization: `Bearer ${process.env.AIMLAPI_KEY}`, "Content-Type": "application/json", }, body: JSON.stringify({ "model": "alibaba/qwen3.7-plus", "messages": [ { "role": "user", "content": "Hello!" } ] }), }); console.log(await r.json());
curl -X POST https://api.aimlapi.com/v1/chat/completions \ -H "Authorization: Bearer $AIMLAPI_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"alibaba/qwen3.7-plus","messages":[{"role":"user","content":"Hello!"}]}'
OpenAI-compatible — swap the base URL and it works with your existing SDK.
| Type | Price |
|---|---|
| Input | |
| Output | |
| Cached input |
| Benchmark | Score | What it measures | Source | Retrieved |
|---|---|---|---|---|
| GPQA Diamond | 90.3% | Google-proof graduate science questions (hardest subset) | Source | July 12, 2026 |
| Terminal-Bench | 70.3% | Autonomous shell/terminal task completion | Source | July 12, 2026 |
| Intelligence | 25.8 | Composite score across standardised reasoning, knowledge and problem-solving evaluations, measured independently by Artificial Analysis | Source | September 12, 2026 |
| Coding | 55.9 | Composite score across standardised coding evaluations, measured independently by Artificial Analysis | Source | September 12, 2026 |
| Model | Input | Output | Context | Best for |
|---|---|---|---|---|
Qwen3.7 Plus This page | Reasoning + agents | |||
| Reasoning + agents | ||||
| Balanced coding + agents | ||||
| Long-context, multimodal & agentic workflows | ||||
| Reasoning + agents |
Qwen3.7 Plus has a 1,000,000 tokens context window and can return up to 65,536 tokens.
Qwen3.7 Plus takes image, text as input and returns text.
Use alibaba/qwen3.7-plus as the model id. Requests go to https://api.aimlapi.com/v1/chat/completions.
Qwen3.7 Plus became available on June 1, 2026.
Qwen3.7 Plus is priced at input $0.416 / 1M tokens, output $1.664 / 1M tokens, cached input $0.0832 / 1M tokens.
Yes, Qwen3.7 Plus can stream responses as they are generated.
Yes, Qwen3.7 Plus accepts image input alongside text.
Qwen3.7 Plus was built by Alibaba Cloud.
No. Its output modality is text only; it accepts image and text input but only generates text.
