Happy Horse 1.0 is now on ModelsLab

Try Now
Skip to main content
Available now on ModelsLab · Language Model

Arcee AI: Trinity Large Preview (free)Frontier MoE Free Now

Try Arcee AI: Trinity Large Preview (free)API Documentation

Scale Efficiently. Infer Fast.

Sparse MoE

400B Total 13B Active

Activates 13B params per token for efficient inference on long contexts up to 512K.

Agentic Workflows

Handles Tools Chains

Navigates complex agent harnesses and multi-step tool actions in production apps.

Creative Chat

Roleplay Storytelling Excels

Delivers strong literacy creativity and character consistency beyond pure reasoning models.

Examples

See what Arcee AI: Trinity Large Preview (free) can create

Copy any prompt below and try it yourself in the playground.

Code Agent

“You are a senior software engineer. Analyze this Python codebase for bugs optimize the sorting algorithm and generate unit tests. Codebase: [insert code]. Output refactored code and tests.”

Long Context Summary

“Summarize this 100K token technical report on sparse MoE architectures. Highlight key innovations in expert routing training efficiency and benchmark results. Extract actionable insights for deployment.”

Creative Story

“Write a 2000-word sci-fi story about an AI agent exploring abandoned megastructures on Mars. Maintain consistent character voices include technical details on sparse MoE systems and build tension across 10 chapters.”

Tool Chain Workflow

“Plan a multi-step workflow to research deploy and monitor a new LLM API. Use tools for web search code generation and logging. Execute steps sequentially with reasoning at each stage.”

For Developers

A few lines of code.
400B MoE. 13B code.

ModelsLab handles the infrastructure: fast inference, auto-scaling, and a developer-friendly API. No GPU management needed.

  • Serverless: scales to zero, scales to millions
  • Pay per token, no minimums
  • Python and JavaScript SDKs, plus REST API
API Documentation 
import requests


response = requests.post(
    "https://modelslab.com/api/v7/llm/chat/completions",
    json={
  "key": "YOUR_API_KEY",
  "prompt": "",
  "model_id": ""
}
)
print(response.json())

FAQ

Common questions about Arcee AI: Trinity Large Preview (free)

Read the docs

It's a 400B-parameter sparse MoE LLM with 13B active per token. Free via OpenRouter preview. Supports long contexts and agentic tasks.

Available on OpenRouter and Arcee API during preview. Use endpoints for chat agents and tools. Deploy weights from Hugging Face.

Trained at 256K supports 512K inference hosted at 128K-256K. Handles extended multi-turn interactions.

Excels in coding logic profound code knowledge and agent harnesses like OpenCode. Matches top open models on benchmarks.

Competes with DeepSeek Qwen GLM in creativity roleplay and agents. Largest U.S. open-weight frontier model.

Built for complex toolchains multiphasic processes and real-time chat. Apache 2.0 license enables production use.

Ready to create?

Start generating with Arcee AI: Trinity Large Preview (free) on ModelsLab.

Try Arcee AI: Trinity Large Preview (free)API Documentation