Build with Qwen API
Access Qwen models via API. Build with QVQ 72B for visual reasoning and Qwen 2.5 72B for speed and efficiency.
QVQ 72B Preview
Advanced visual reasoning
Built upon Qwen2-VL-72B, QVQ is perfect for solving complex problems that require combined text and visual reasoning, such as interpreting graphs, diagrams, and equations in educational and scientific contexts.
Potential use cases include visual question-answering systems, interactive learning platforms, and automated content generation.
How to use QVQ-72B-Preview via API.
Features and Performance
Superior Mathematical Analysis
Surpasses Qwen2-VL-72B-Instruct on benchmarks like MathVista and MathVision, excelling in algebraic reasoning, function graphs, and scientific problem-solving.
Exceptional Multimodal Reasoning
QVQ-72B-Preview achieves a leading score of 70.3 on MMMU, narrowing the gap with top-tier models like GPT o1 and GPT 4o.
Advanced Visual and Logical Reasoning
QVQ-72B-Preview demonstrates precision, focus, and analytical depth in image-content-based and logic-heavy tasks, closely aligning with Claude 3.5 Sonnet.
Qwen2.5 72B Instruct Turbo
Well-balanced intelligence and speed
Qwen2.5 72B Instruct Turbo handles up to 32K tokens. Ideal for scenarios prioritizing complexity as well a speed and computational efficiency, like coding tasks or real-time chatbot applications, which don’t need extended context.
How to use Qwen2.5-72B-Instruct-Turbo via API.
Features and Performance
Rapid Coding Assistance
Qwen2.5-72B-Instruct-Turbo excels in debugging, generating code snippets, and producing structured outputs with scores of 86.6 on HumanEval and 88.2 on MBPP.
Precise Instruction Following
Qwen2.5-72B-Instruct Turbo delivers precise, coherent responses with a 75 Quality Index (close to Deepseek V3), excelling in role-play, system prompts, and structured output tasks.
High Speed Output
Qwen2.5-72B-Instruct Turbo outpaces its standard counterpart, facilitating real-time decision-making and dynamic workflows for time-critical applications.
Qwen 2.5 7B Instruct Turbo
Fast and resource-efficient performance
Qwen2.5 7B Instruct Turbo is a highly optimized, instruction-tuned large language model designed for coding, multilingual tasks, and structured data understanding. Compared to the 72B model, it offers the benefits of greater efficiency, lower computational requirements, and faster response times.
How to use Qwen 2.5 7B Instruct Turbo via API.
Features and Performance
Coding and Mathematics
Qwen2.5-7B-Instruct-Turbo outperforms previous iterations in domain-specific tasks, enhancing precision and utility for technical users with 49.8 MATH and 57.9 HumanEval.
Structured Inputs and Outputs
Qwen2.5-7B-Instruct-Turbo understands structured inputs like tables and provides reliable generation of structured formats, especially JSON.
Scalability for Mid-range
Optimized for medium-scale use cases, making it a better fit for developers and startups seeking advanced NLP without excessive overhead.
Qwen 2 72B Instruct
Context handling and multilingual support
Building on the strengths of Qwen 1.5, Qwen 2 Instruct (72B) advances NLP to engage users in dynamic and interactive conversations thanks to expanded linguistic capabilities. It incorporates advanced features like SwiGLU activation and group query attention for improved efficiency in processing information.
How to use Qwen 2 72B Instruct via API.
Features and Performance
Benchmark Performance
Qwen 2 72B Instruct achieves scores of 64.4 on MMLU-Pro and 86.0 on HumanEval, outperforming Llama-3-70B-Instruct.
Advanced Multilingual Capabilities
Trained on data from 27 languages, demonstrating exceptional performance in Asian languages compared to earlier models.
Transformer Architecture
Qwen 2 72B Instruct features SwiGLU, QKV bias, group query, and sliding window attention for better efficiency and focus.
Developing with Qwen API
Gain instant access to Qwen 2.5, Qwen 2, and QVQ 72B through production API: AI/ML API provides direct connectivity to Qwen.
Try Playground Qwen 1.5 API for free.
100+ AI Models
Cost-Effective API
High Performance
Serverless Architecture
import os
from openai import OpenAI
client = OpenAI(
base_url="<YOUR_API_KEY>",
api_key="https://api.aimlapi.com/v1",
)
response = client.chat.completions.create(
model="qwen/qvq-72b-preview",
messages=[
{
"role": "system",
"content": "You are an AI assistant who knows everything.",
},
{
"role": "user",
"content": "Tell me, why is the sky blue?"
},
],
)
message = response.choices[0].message.content
print(f"Assistant: {message}")How to Get an Qwen API Key with AI/ML API
Sign Up
Begin by signing up on our AI/ML API platform. Create your account to gain access to a wide range of powerful AI models and tools.
Create API Key
In the Playground, navigate to the Key Management section and click on Create API Key. You can easily activate or deactivate your keys as needed.
Integrate AI
After creating your API key, you can integrate AI models into your application by following the guidelines provided in our API reference.
