qwen3-next-80b-a3b-instruct

qwen3-next-80b-a3b-instruct

Efficient Mixture-of-Experts (MoE) Instruction-Finetuned Model
2025-09-11
LLM
Input:
$0.143/1M tokens
Output:
$0.5715/1M tokens
Bulk order? Contact your manager for exclusive deals

API Overview

Qwen3-Next-80B-A3B-Instruct is the next-generation, highly efficient Mixture-of-Experts (MoE) instruction-tuned model released by Alibaba’s Tongyi Lab. Its core positioning is as a mainstream general-purpose large model that offers “high cost-effectiveness, strong instruction following, and enterprise-grade usability.”

  • Exquisite MoE Architecture: With a total of 80 billion parameters, it activates only 3 billion parameters (A3B), significantly reducing inference costs and latency while maintaining powerful language capabilities.
  • Outstanding Instruction-Following Ability: After meticulous alignment training, it delivers stable and reliable performance in complex instruction tasks such as multi-turn dialogues, format constraints, and style control.
  • Ultra-long Context Support up to 128K Tokens: It can handle long documents, multi-turn historical dialogues, or composite task inputs with high recall of critical information.
  • Multi-language and Code Optimization: Deeply enhanced understanding of Chinese contexts, while also supporting mainstream languages such as English, Japanese, and French, as well as programming languages like Python and JavaScript.

───────────────────────────────────────────────────────────────────

Core Capabilities

High Throughput and Low Latency Response: With few activated parameters and low GPU memory usage, a single card can support highly concurrent enterprise-level applications such as customer service, content generation, and intelligent office automation.

🧠 Precise Task Execution: It accurately understands fine-grained instructions such as “summarize in table form,” “write in official government document style,” or “list pros and cons in bullet points.”

🌍 Natural Bilingual Expression in Chinese and English: The output is authentic and fluent, balancing technical terminology with cultural context, making it suitable for both international business and localized scenarios.

🛡️ Secure, Compliant, and Controllable: Supports content filtering, sensitive word interception, and audit logs, meeting deployment requirements in industries such as finance, government, and education.

Playground

Log in to explore more features! Click to Log In

API Analytics

API Reference (1)

API DescriptionAPI EndpointRequest MethodStabilityParameter Description
qwen3-next-80b-a3b-instruct
POST
Stable
View Details

API Pricing

$
ModelDescriptionContextOfficial Price302.AI Price

qwen3-next-80b-a3b-instruct

-
128000

Input$0.143 / 1M tokens
Output$0.5715 / 1M tokens

Input$0.143/ 1M tokens
Output$0.5715/ 1M tokens
Original Price