Model Information

Display Name: KAT-Coder-Air V2.5

API Model ID: kwaipilot/kat-coder-air-v2.5

Category: Text To Text

Description: KAT-Coder-Air V2.5 is Kwaipilot's agentic coding model: you can hand it an entire issue or an entire business workflow and it autonomously locates the relevant code, makes the modifications and carries the task through to completion directly in the repository. V2.5 integrates multiple domain experts while fully retaining the front-end aesthetic generation strength of V2. **Key Features:** - 256,000-token context window (large-codebase work) - Up to 80K output tokens - Agentic coding: autonomous, multi-step issue resolution - Function/tool calling for repository and workflow automation - Always-on internal reasoning (counted as output tokens) - Front-end / UI generation quality inherited from V2 - Cache-read pricing for repeated context **Best For:** - End-to-end issue resolution in real repositories - Multi-file refactoring and agentic coding loops - Front-end and UI generation - Cost-sensitive coding at scale (Air tier)

Context Window: 256,000 tokens

Max Output: 80,000 tokens

How to Use This Model

To use KAT-Coder-Air V2.5 via the HInow.ai API, use the model ID: kwaipilot/kat-coder-air-v2.5

API Request Example (Chat/Text)

POST https://api.hinow.ai/v1/chat/completions
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

{
  "model": "kwaipilot/kat-coder-air-v2.5",
  "messages": [
    {"role": "user", "content": "Your message here"}
  ]
}
              

Pricing

• input: $0.2025
• output: $0.81
• cached: $0.0405

Available Parameters

• temperature: Controls randomness (0-2). Default: 0.3 (Options: 0, 0.2, 0.3, 0.5, 0.7, 1.0)
• top_p: Nucleus sampling (0-1). Default: 0.95 (Options: 0.1, 0.5, 0.8, 0.9, 0.95, 1.0)
• max_tokens: Max tokens to generate, including internal reasoning (1-80000). Default: 8192 (Options: 512, 1024, 2048, 4096, 8192, 16384, 32768, 80000)
• tools: Function/tool definitions for agentic coding workflows
• response_format: Output format (structured outputs supported) (Options: text, json_object)
• stop: Sequences where the API stops generating further tokens
• frequency_penalty: Penalizes token frequency (-2 to 2). Default: 0 (Options: 0, 0.5, 1.0)
• presence_penalty: Penalizes repeated topics (-2 to 2). Default: 0 (Options: 0, 0.5, 1.0)

Quick Reference

To use this model, set: "model": "kwaipilot/kat-coder-air-v2.5"

Featured: No

Documentation: https://hinow.ai/models/kwaipilot/kat-coder-air-v2.5

API Endpoint: https://api.hinow.ai/v1