Models

Public model

AtlasFlux exposes a single model identifier:

atlasflux/nenas-flash

You do not need to specify model names. AtlasFlux automatically routes each request to the best model.

Model capabilities

Capability	Supported
Text generation	Yes
Vision (images)	Yes
Reasoning	Yes
Web search	Yes
Streaming	Yes
Tool calling	Yes
Structured output (JSON)	Yes
Context window	1,000,000 tokens
Max output	32,000 tokens

Intelligent routing

When you send a request, AtlasFlux evaluates:

Task type — Code, analysis, creative writing, etc.
Complexity — Simple questions vs. multi-step reasoning
Input features — Images, tools, structured output requirements
Reasoning effort — Whether reasoning mode is requested

Based on this, the router selects the optimal model from its internal model pool.

Routing categories

Category	Description
`general`	Standard text tasks, Q&A, summarization
`multimodal`	Tasks involving images or audio
`reasoning`	Complex analysis, math, multi-step problems
`structured`	JSON schema, function calling, structured output
`classifier`	Simple classification and categorization tasks

Model response

The response includes routing metadata:

{
  "routing": {
    "category": "general",
    "upstream_model": "atlasflux/nenas-flash",
    "reasoning": "Selected for general text generation"
  }
}

Pricing

All models use the same AtlasFlux pricing:

Token Type	Rate per 1M tokens
Input	RM 5.00
Output	RM 25.00
Reasoning	RM 25.00

Prices are in MYR and deducted from your prepaid wallet balance.

Core Concepts

Guides

API Reference

Resources

Public model

Model capabilities

Intelligent routing

Routing categories

Model response

Pricing

​Public model

​Model capabilities

​Intelligent routing

​Routing categories

​Model response

​Pricing

Public model

Model capabilities

Intelligent routing

Routing categories

Model response

Pricing