Qwen3.8-Max Added and DeepSeek-V4-Flash Upgraded

Date: 2026-08-03 / (1405-05-12)

Summary

Alibaba's new qwen3.8-max flagship is now available on AvalAI for long-horizon coding, professional work, multimodal understanding, and agentic workflows. The existing deepseek-v4-flash model ID now automatically routes to DeepSeek-V4-Flash-0731, the official release with stronger agentic capabilities; no application changes are required and pricing is unchanged.

---

Details

Alibaba: Qwen3.8-Max

We announce access to Qwen3.8-Max (qwen3.8-max), Alibaba's new 2.4-trillion-parameter mixture-of-experts flagship. It is designed to plan, execute, and verify long-running coding and professional workflows while understanding text, images, video, and very large documents. Alibaba models documentation

Key Features:

• Long-Horizon Coding: Designed to autonomously implement and deliver multi-stage projects that can span more than 10 days
• Professional Work: Supports end-to-end tasks across legal, financial, design, and other specialist domains
• Native Multimodality: Accepts text, image, and video input and returns text
• 1M-Token Context: Up to 991,000 input tokens for large repositories, document collections, and extended video analysis
• 128K Output Capacity: Up to 128,000 output tokens for substantial code, analysis, and deliverables
• Deep Thinking and Tools: Supports reasoning, function calling, tool choice, structured output, prompt caching, and web search
• Endpoint Support: Full support on v1/chat/completions and v1/messages; partial support on v1/responses

DeepSeek: Automatic V4 Flash Upgrade

The AvalAI model ID deepseek-v4-flash now resolves to DeepSeek-V4-Flash-0731, DeepSeek's official V4 Flash release. This release supersedes the preview and substantially improves long-horizon coding, terminal work, tool use, automation, and other agentic workloads.

• No code change required: Continue sending "model": "deepseek-v4-flash"
• Pricing unchanged: Input, cached-input, and output rates remain the same
• Model structure: 284B total parameters with 13B active parameters and an attached DSpark speculative-decoding module
• Reasoning effort: Supports low, high, and max
• Context window: 1 million tokens
• Official release: The 0731 checkpoint replaces the earlier preview behind the stable AvalAI model ID

Endpoint Availability

Model	v1/chat/completions	v1/messages	v1/responses
qwen3.8-max	Full	Full	Partial
deepseek-v4-flash	Full	—	—

---

Pricing

Prices are in USD per 1 million tokens.

Model	Input	Cached Input	Cache Creation Input	Output
qwen3.8-max	$2.00	$0.25	$2.50	$6.00
deepseek-v4-flash	$0.14	$0.0028	—	$0.28

The DeepSeek upgrade does not change the existing deepseek-v4-flash rates.

---

API Request and Response Example

Request

curl https://api.avalai.ir/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "qwen3.8-max",
    "messages": [
      {
        "role": "user",
        "content": "Review this platform architecture and propose a staged implementation plan with verification gates."
      }
    ],
    "extra_body": {"enable_thinking": false}
  }'

Response

The following shortened response illustrates the standard Chat Completions structure:

{
  "id": "chatcmpl-qwen38-example",
  "created": 1785744000,
  "model": "qwen3.8-max",
  "object": "chat.completion",
  "system_fingerprint": null,
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "content": "Start by defining stable service boundaries and measurable rollback criteria. Introduce observability and contract tests before migrating one low-risk workflow behind a feature flag, then expand traffic only after each verification gate passes.",
        "role": "assistant",
        "thinking_blocks": [],
        "annotations": []
      }
    }
  ],
  "usage": {
    "completion_tokens": 55,
    "prompt_tokens": 23,
    "total_tokens": 78,
    "completion_tokens_details": null,
    "prompt_tokens_details": {
      "audio_tokens": null,
      "cached_tokens": 0,
      "text_tokens": 23,
      "image_tokens": null
    }
  },
  "estimated_cost": {
    "unit": "0.0003760000",
    "irt": 57.6,
    "exchange_rate": 153200
  }
}

---

SDK Usage Examples

Bash

curl https://api.avalai.ir/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "qwen3.8-max",
    "messages": [
      {
        "role": "user",
        "content": "Design a reliable implementation plan for this multi-service platform."
      }
    ],
    "extra_body": {"enable_thinking": false}
  }'

Python

from openai import OpenAI

client = OpenAI(
    api_key="your-avalai-api-key",
    base_url="https://api.avalai.ir/v1",
)

response = client.chat.completions.create(
    model="qwen3.8-max",
    messages=[
        {
            "role": "user",
            "content": "Design a reliable implementation plan for this multi-service platform.",
        }
    ],
    extra_body={"enable_thinking": False},
)

print(response.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 response = await client.chat.completions.create({
  model: "qwen3.8-max",
  messages: [
    {
      role: "user",
      content: "Design a reliable implementation plan for this multi-service platform.",
    },
  ],
  extra_body: { enable_thinking: false },
});

console.log(response.choices[0].message.content);

Continue Using DeepSeek-V4-Flash Without Changes

curl https://api.avalai.ir/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [
      {
        "role": "user",
        "content": "Find the root cause of this failing deployment and propose the safest fix."
      }
    ],
    "reasoning_effort": "high"
  }'

The same stable model ID now uses DeepSeek-V4-Flash-0731 automatically.

---

Related Links

• Qwen3.8-Max Model Details
• Alibaba Models Overview
• DeepSeek Models Overview
• Reasoning Models Guide
• Chat Completions API Reference
• Pricing