Ask AI

New Models Added: Qwen3.5 Plus, Qwen3.5-397B and MiniMax M2.5

Date: 2026-02-17 / (1404-11-28)

Summary

AvalAI introduces four new AI models: Qwen3.5 Plus and Qwen3.5-397B-A17B from Alibaba, featuring hybrid architecture with linear attention and sparse mixture-of-experts for enhanced multimodal understanding; and MiniMax M2.5 with M2.5-Lightning, SOTA models for coding, agentic tool use, and real-world productivity tasks with unprecedented cost efficiency.

Details

Alibaba

We announce access to Qwen3.5 Plus (qwen3.5-plus) and Qwen3.5-397B-A17B (qwen3.5-397b-a17b), Alibaba's latest native vision-language models built on innovative hybrid architecture. Documentation

qwen3.5-plus

Key Features:

Native Vision-Language Model: Built for multimodal understanding with text, image, and video inputs
Hybrid Architecture: Integrates linear attention (via Gated Delta Networks) with sparse mixture-of-experts for higher inference efficiency
1M Context Window: Extended context available via Alibaba Cloud Model Studio
Deep Thinking: Advanced reasoning capabilities on par with leading frontier models
Official Built-in Tools: Adaptive tool use with official tool support
Function Calling & Structured Output: Full support for advanced API features
201 Languages: Expanded language and dialect support from 119 to 201 languages
Endpoint Support: Available on v1/chat/completions

Feature	Details
Model ID	`qwen3.5-plus`
Snapshot	`qwen3.5-plus-2026-02-15`
Context Window	1,000,000 tokens
Input	Text, Image, Video
Output	Text
Input Pricing	$0.40 / 1M tokens
Input Above 256K	$1.20 / 1M tokens
Cache Creation	$0.50 / 1M tokens
Cache Creation Above 256K	$1.50 / 1M tokens
Cached Input	$0.04 / 1M tokens
Cached Input Above 256K	$0.12 / 1M tokens
Output Pricing	$2.40 / 1M tokens
Output Above 256K	$7.20 / 1M tokens

qwen3.5-397b-a17b

Key Features:

397B Total Parameters: Massive model with only 17B activated per forward pass for efficiency
Native Vision-Language Model: Built for multimodal understanding with text, image, and video inputs
State-of-the-Art Performance: Comparable to leading-edge models across language understanding, logical reasoning, code generation, agent-based tasks, image understanding, video understanding, and GUI interactions
Open-Weight Model: Available for download on Hugging Face and ModelScope
Strong Agent Capabilities: Robust code-generation and agent capabilities with excellent generalization
Deep Thinking: Advanced reasoning and visual understanding
Endpoint Support: Available on v1/chat/completions

Feature	Details
Model ID	`qwen3.5-397b-a17b`
Parameters	397B total, 17B activated
Input	Text, Image, Video
Output	Text
Input Pricing	$0.60 / 1M tokens
Cached Input	$0.06 / 1M tokens
Output Pricing	$3.60 / 1M tokens

Use Cases:

Complex multimodal reasoning and analysis
Long document and video understanding
Agent-based tasks and autonomous workflows
GUI interaction and visual understanding
Multilingual applications across 201 languages
Code generation and software engineering

MiniMax

We announce access to MiniMax M2.5 (minimax-m2.5) and MiniMax M2.5-Lightning (minimax-m2.5-lightning), MiniMax's latest models designed for real-world productivity with SOTA performance in coding and agentic tasks. Documentation

minimax-m2.5

Key Features:

SOTA Coding Performance: 80.2% on SWE-Bench Verified, 51.3% on Multi-SWE-Bench
Advanced Search & Tool Calling: 76.3% on BrowseComp with context management
37% Faster Than M2.1: Improved task decomposition and token efficiency
Spec-Writing Architecture: Model actively plans features, structure, and UI design before coding
Built-in Reasoning: Thinking content in <think> tags, separable via reasoning_split parameter
10+ Programming Languages: Go, C, C++, TypeScript, Rust, Kotlin, Python, Java, JavaScript, PHP, Lua, Dart, Ruby
Full Development Lifecycle: From 0-to-1 system design to 90-to-100 code review and testing
Office Work Integration: Deep collaboration with finance, law, and social sciences professionals
Cost Effective: $1 for 1 hour continuous operation at 100 tokens/second
Endpoint Support: Available on v1/chat/completions

Feature	Details
Model ID	`minimax-m2.5`
Context Window	204,000 tokens
Output Speed	~50 tokens per second
Input Pricing	$0.30 / 1M tokens
Cached Input	$0.03 / 1M tokens
Cache Creation	$0.375 / 1M tokens
Output Pricing	$1.20 / 1M tokens

minimax-m2.5-lightning

Key Features:

Ultra-Fast Inference: ~100 tokens per second (2x faster than other frontier models)
Same Capabilities as M2.5: Identical intelligence with higher throughput
On Par with Claude Opus 4.6: Matches completion speed of leading models
Low-Cost Operation: $0.30 per hour at 100 TPS continuous operation
Full Caching Support: Context caching for cost optimization
Endpoint Support: Available on v1/chat/completions

Feature	Details
Model ID	`minimax-m2.5-lightning`
Context Window	204,000 tokens
Output Speed	~100 tokens per second
Input Pricing	$0.30 / 1M tokens
Cached Input	$0.03 / 1M tokens
Cache Creation	$0.375 / 1M tokens
Output Pricing	$2.40 / 1M tokens

Use Cases:

Agentic coding and software engineering (SWE-Bench Verified 80.2%)
Multi-programming language projects
Complex search and research tasks
Office work automation (Word, PowerPoint, Excel)
Financial modeling and analysis
Real-time code assistance

Pricing Summary

Model	Provider	Type	Pricing
`qwen3.5-plus`	Alibaba	Chat	Input: $0.40/1M, Cached: $0.04/1M, Output: $2.40/1M (tiered above 256K)
`qwen3.5-397b-a17b`	Alibaba	Chat	Input: $0.60/1M, Cached: $0.06/1M, Output: $3.60/1M
`minimax-m2.5`	MiniMax	Chat	Input: $0.30/1M, Cached: $0.03/1M, Output: $1.20/1M
`minimax-m2.5-lightning`	MiniMax	Chat	Input: $0.30/1M, Cached: $0.03/1M, Output: $2.40/1M

API Request/Response Examples

Qwen3.5 Plus Chat Example

Request

bash

curl https://api.avalai.ir/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "qwen3.5-plus",
    "messages": [
      {
        "role": "user",
        "content": "Design an intelligent agent system that can autonomously manage complex multi-step workflows."
      }
    ],
    "max_tokens": 4096,
    "stream": false,
    "extra_body": {"enable_thinking": false}
  }'

Response

json

{
  "id": "chatcmpl-xyz123",
  "created": 1739794897,
  "model": "qwen3.5-plus",
  "object": "chat.completion",
  "system_fingerprint": null,
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "content": "Here's a comprehensive design for an intelligent agent system...",
        "role": "assistant",
        "thinking_blocks": [],
        "annotations": []
      }
    }
  ],
  "usage": {
    "completion_tokens": 1024,
    "prompt_tokens": 25,
    "total_tokens": 1049
  },
  "estimated_cost": {
    "unit": "0.0024596",
    "irt": 323.11,
    "exchange_rate": 131350
  }
}

MiniMax M2.5 Coding Example

Request

bash

curl https://api.avalai.ir/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "minimax-m2.5",
    "messages": [
      {
        "role": "user",
        "content": "Implement a rate-limited concurrent web crawler in Rust with proper error handling."
      }
    ],
    "max_tokens": 4096
  }'

Response

json

{
  "id": "chatcmpl-abc456",
  "created": 1739794897,
  "model": "minimax-m2.5",
  "object": "chat.completion",
  "system_fingerprint": null,
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "content": "Here's a complete implementation of a rate-limited concurrent web crawler in Rust...",
        "role": "assistant",
        "thinking_blocks": [],
        "annotations": []
      }
    }
  ],
  "usage": {
    "completion_tokens": 2048,
    "prompt_tokens": 30,
    "total_tokens": 2078
  },
  "estimated_cost": {
    "unit": "0.0024666",
    "irt": 324.03,
    "exchange_rate": 131350
  }
}

MiniMax M2.5-Lightning Fast Response Example

Request

bash

curl https://api.avalai.ir/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "minimax-m2.5-lightning",
    "messages": [
      {
        "role": "user",
        "content": "Explain the difference between async/await and Promises in JavaScript."
      }
    ]
  }'

Response

json

{
  "id": "chatcmpl-def789",
  "created": 1739794897,
  "model": "minimax-m2.5-lightning",
  "object": "chat.completion",
  "system_fingerprint": null,
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "content": "The key differences between async/await and Promises in JavaScript are...",
        "role": "assistant",
        "thinking_blocks": [],
        "annotations": []
      }
    }
  ],
  "usage": {
    "completion_tokens": 512,
    "prompt_tokens": 18,
    "total_tokens": 530
  },
  "estimated_cost": {
    "unit": "0.0012339",
    "irt": 162.08,
    "exchange_rate": 131350
  }
}

MiniMax M2.5 with Reasoning Split

Use reasoning_split to separate the model's thinking process from the final answer:

Request

bash

curl https://api.avalai.ir/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $AVALAI_API_KEY" \
  -d '{
    "model": "minimax-m2.5",
    "messages": [
      {
        "role": "user",
        "content": "What is 25 * 37?"
      }
    ],
    "extra_body": {
      "reasoning_split": true
    }
  }'

Response

json

{
  "id": "chatcmpl-xyz123",
  "created": 1739794897,
  "model": "minimax-m2.5",
  "object": "chat.completion",
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "content": "The answer is 925.",
        "role": "assistant",
        "reasoning_details": "Let me calculate 25 * 37 step by step:\n25 * 37 = 25 * (30 + 7) = 750 + 175 = 925"
      }
    }
  ],
  "usage": {
    "completion_tokens": 45,
    "prompt_tokens": 12,
    "total_tokens": 57
  }
}

Note

Without reasoning_split, the thinking content appears inline with <think> tags in the content field. When managing multi-turn conversations with tool calls, always preserve the full response message (including tool_calls and reasoning_details fields) in your message history.

New Models Added: Qwen3.5 Plus, Qwen3.5-397B and MiniMax M2.5 ​

Summary ​

Details ​

Alibaba ​

qwen3.5-plus ​

qwen3.5-397b-a17b ​

MiniMax ​

minimax-m2.5 ​

minimax-m2.5-lightning ​

Pricing Summary ​

API Request/Response Examples ​

Qwen3.5 Plus Chat Example ​

Request ​

Response ​

MiniMax M2.5 Coding Example ​

Request ​

Response ​

MiniMax M2.5-Lightning Fast Response Example ​

Request ​

Response ​

MiniMax M2.5 with Reasoning Split ​

Request ​

Response ​

Related Documentation ​

New Models Added: Qwen3.5 Plus, Qwen3.5-397B and MiniMax M2.5

Summary

Details

Alibaba

qwen3.5-plus

qwen3.5-397b-a17b

MiniMax

minimax-m2.5

minimax-m2.5-lightning

Pricing Summary

API Request/Response Examples

Qwen3.5 Plus Chat Example

Request

Response

MiniMax M2.5 Coding Example

Request

Response

MiniMax M2.5-Lightning Fast Response Example

Request

Response

MiniMax M2.5 with Reasoning Split

Request

Response

Related Documentation