Poolside Laguna M.1 and Laguna XS.2 Are Now Available in Puter.js

Puter.js now supports Laguna M.1 and Laguna XS.2, Poolside's first two foundation models — both purpose-built for agentic coding. Add them to your application for free, no API keys or Poolside account required.

What is Laguna M.1?

Laguna M.1 is Poolside's flagship agentic coding model, trained from scratch in-house on 30T tokens across 6,144 NVIDIA Hopper GPUs. It's a Mixture-of-Experts architecture with 225B total parameters and 23B activated per token, designed for complex, long-horizon software engineering tasks. Key highlights include:

• 128K token context window with native tool calling and reasoning
• 72.5% on SWE-bench Verified and 46.9% on SWE-bench Pro, placing it in the same tier as far larger frontier models
• Sparse MoE inference — frontier-tier coding capability at the active compute cost of a much smaller dense model
• Built for agentic workflows: writing code, running tests, inspecting failures, and iterating across files end to end

What is Laguna XS.2?

Laguna XS.2 is Poolside's second-generation MoE and their first open-weight release, available under the Apache 2.0 license. At 33B total parameters with 3B activated per token, it's compact enough to run locally on a Mac with 36 GB of RAM while still delivering frontier-adjacent coding performance. Key highlights include:

• 68.2% on SWE-bench Verified, 62.4% on SWE-bench Multilingual, 44.5% on SWE-bench Pro, and 30.1% on Terminal-Bench 2.0
• Apache 2.0 open weights — permissive for commercial use, modification, and redistribution
• 3B activated parameters for low inference cost and fast local execution
• Incorporates lessons from the M.1 training pipeline, including refined synthetic data and RL stages

Examples

Agentic coding with Laguna M.1

puter.ai.chat("Refactor this Express route handler into a typed tRPC procedure with input validation, then write integration tests covering the error paths", {
    model: "poolside/laguna-m.1:free"
})

Multi-step engineering with the flagship

puter.ai.chat("Trace how a webhook flows through the queue worker into the database, identify any places retries could double-write, and propose a fix", {
    model: "poolside/laguna-m.1:free"
})

Local-friendly coding with Laguna XS.2

puter.ai.chat("Write a Python script that walks a monorepo, finds all package.json files, and generates a dependency graph as a Mermaid diagram", {
    model: "poolside/laguna-xs.2:free"
})

Streaming with reasoning

const response = await puter.ai.chat(
    "Design a leader-election protocol for a 5-node cluster that tolerates network partitions, and walk through the trade-offs",
    { model: "poolside/laguna-m.1:free", stream: true }
);

for await (const part of response) {
    if (part?.reasoning) puter.print(part?.reasoning);
    else puter.print(part?.text);
}

Get Started Now

Just add one library to your project:

// npm install @heyputer/puter.js
import { puter } from '@heyputer/puter.js';

Or add one script tag to your HTML:

<script src="https://js.puter.com/v2/"></script>

No API keys and no infrastructure setup. Start building with Laguna M.1 and Laguna XS.2 immediately.

Learn more:

• Poolside Models on Puter.js
• Puter.js AI Documentation