What makes these models different from GitHub Copilot or Cursor?

Poolside provides open-weight foundation models specifically trained for agentic, multi-step coding tasks with tool use, not just single-line autocomplete. You get full control over the model and its environment.

How does the 1M token context window work in practice?

It allows the agent to ingest entire large codebases into its working memory for coherent analysis, refactoring, or debugging across many files without losing context.

What is the 'Active' parameter count (e.g., 8B active in Laguna S)?

This refers to a Mixture-of-Experts (MoE) architecture. While the total parameter count is 118B, only a subset (8B) are activated for any given token, dramatically improving inference speed and efficiency.

How do I configure custom tools for the agent?

Use the skills and MCP server configuration system defined in your AGENTS.md or via CLI settings. This allows you to expose arbitrary shell commands, APIs, or scripts as tools the agent can invoke.

Directory/Coding/AI Assistants/Poolside

Poolside

Poolside builds and provides open-weight foundation models optimized for agentic coding tasks. Their flagship model, Laguna S 2.1, is a frontier-class reasoning model with 118B parameters designed for quality and efficiency. A lighter variant, Laguna XS 2.1, with 33B parameters is small enough for on-device use. The platform offers a CLI agent, API access, and integrations with tools like OpenRouter and Vercel AI Gateway, targeting developers building software with AI assistance.

Coding

Visit website

Added on

Sep 19, 2026

Open-weight modelsAgentic codingCLI agent1M context windowOn-device optionResearch-backed

Product information

Everything worth knowing about Poolside

The complete picture — from everyday features to the technical detail builders and IT folks dig for.

Core features

What it actually does

Agentic Coding Models

Access open-weight foundation models (Laguna S 2.1, Laguna XS 2.1) optimized for code generation, reasoning, and long-context tasks.

Terminal/CLI Agent

Use the 'pool' CLI to interact with Poolside models directly from your terminal for coding tasks, file edits, and automation.

On-Device Model Deployment

Run the lightweight Laguna XS model locally on your device for fast, private coding assistance.

Tool & Editor Integration

Integrate Poolside models into your existing development workflow via desktop apps, CLI, and supported editors.

Technical capabilities

Open-Weight & Research-Driven

Models are built on published research with details on training, evaluation, and architecture (e.g., 118B params, 1M context window for Laguna S).

Must watch videos

Who uses it

Real use cases, no hype

Autonomous Code Refactoring

Use Laguna S's massive context to audit, plan, and execute large-scale code migrations across entire repositories.

Local Development Assistant

Run Laguna XS on-device for instant, offline code completion and chat without sending code to the cloud.

Secure Enterprise Agent Deployment

Leverage sandboxes and fine-grained permissions to run untrusted agent scripts in isolated environments.

Multi-Step API Automation

Chain model calls with tool integrations (like web search, file I/O) to build complex automated pipelines.

Research & Fine-Tuning

Access open weights to conduct research, run ablations, or fine-tune on proprietary codebases.

Interactive CLI Debugging

Use the interactive terminal mode to iteratively debug issues by asking natural language questions about your code.

What's great

Open-weight models: full access to weights for customization.
Two distinct model sizes: Laguna S (118B) for quality, XS (33B) for speed/cost.
Massive context windows: up to 1M tokens for long-horizon tasks.
Agentic coding focus: built specifically for multi-step code generation and tool use.

Technical strengths

Local deployment possible: XS model runs on-device, reducing latency and costs.
Native tool integrations and subagent support for complex workflows.
Fine-grained permission and sandbox system for secure agent execution.

Where it falls short

Newer ecosystem: less community tooling and fine-tunes compared to giants like Llama.
Steep learning curve for full CLI and configuration system.
Complex pricing: separate tiers for on-device vs. cloud usage.

Technical limitations

API-only for cloud access; requires separate setup for self-hosting.
Limited documentation depth for advanced MLOps and custom training.

For technical folks

The deep-dive specs

FAQ

Includes technical Q&A

Yes. The Laguna XS 2.1 model is small enough (33B params, 3B active) to run on-device. Laguna S requires more significant GPU resources but the weights are open for self-hosting.

Traffic Insights

Monthly Visits

—

October 2026

Growth Rate

vs last month

Dominance

in Coding

Top Country

—

Top Countries Breakdown

—

Growth Trend

Stable

Traffic has remained stable.

Pricing overview

Freemium Pricing

Poolside Pricing

Poolside uses a freemium pricing structure. Real-time rates, custom enterprise quotas, and plan details are managed directly on their official platform.

Check Pricing on Poolside

Reviews