New Model Added: Qwen3-Max-Preview with Enhanced Performance and Stability Improvements
Date: 2025-09-09
Summary
We're excited to announce the addition of Qwen3-Max-Preview, Alibaba's most powerful large language model with over 1 trillion parameters, now available on AvalAI. Additionally, we've implemented significant service stability improvements to ensure better performance during peak usage hours.
Details
Alibaba Qwen3-Max-Preview
We've added Alibaba's flagship Qwen3-Max-Preview model, featuring breakthrough capabilities and performance:
- qwen3-max-preview: Alibaba's largest model with 1+ trillion parameters, offering exceptional reasoning capabilities, advanced coding support, and superior performance across complex tasks. Documentation
Key Features:
- Context Window: 262,144 tokens (258,048 input, 32,768 output)
- Advanced Reasoning: Superior performance on complex analytical tasks
- Coding Excellence: Enhanced programming and code generation capabilities
- Multimodal Support: Handles structured data formats like JSON
- Context Caching: Optimized performance for extended sessions
API Endpoints:
- Primary Support:
v1/chat/completions(full feature compatibility) - Partial Support:
v1/messages(basic text generation)
python
from openai import OpenAI
client = OpenAI(api_key="your-avalai-api-key", base_url="https://api.avalai.ir/v1")
completion = client.chat.completions.create(
model="qwen3-max-preview",
messages=[
{
"role": "user",
"content": "Analyze the economic implications of AI adoption across different industries.",
}
],
max_tokens=1000,
)
print(completion.choices[0].message.content)javascript
import { OpenAI } from "openai";
const client = new OpenAI({
apiKey: process.env.AVALAI_API_KEY,
baseURL: "https://api.avalai.ir/v1",
});
const completion = await client.chat.completions.create({
model: "qwen3-max-preview",
messages: [
{
role: "user",
content: "Analyze the economic implications of AI adoption across different industries.",
},
],
max_tokens: 1000,
});
console.log(completion.choices[0].message.content);Service Stability Improvements
We've implemented comprehensive infrastructure enhancements to improve service reliability:
Enhanced Performance:
- Load Balancing: Improved distribution of requests across our infrastructure
- Peak Hour Optimization: Better resource allocation during high-traffic periods
- Response Time: Reduced latency for all model endpoints
- Error Handling: Enhanced fault tolerance and recovery mechanisms
User Benefits:
- More consistent response times during peak usage
- Reduced service interruptions
- Improved overall API reliability
- Better handling of concurrent requests
Benchmark Performance
Qwen3-Max-Preview demonstrates exceptional performance across industry benchmarks:
- SuperGPQA: Leading performance in advanced reasoning
- AIME25: Superior mathematical problem-solving capabilities
- LiveCodeBench v6: Outstanding coding and programming tasks
- Arena-Hard v2: Excellent general-purpose performance
- LiveBench: Consistent top-tier results across diverse tasks