Back to News
PoolsideJuly 2, 2026

Laguna XS 2.1

Open Source
LLM
Coding Agent
Open Weights
Mixture of Experts

Explore Laguna XS 2.1

Visit the official website to learn more and get started

### TL;DR

Laguna XS 2.1 is a 33B total parameter Mixture-of-Experts (MoE) coding agent model with 3B activated parameters per token, specifically engineered for high-performance agentic coding and long-horizon tasks. It features an efficient architecture with mixed sliding-window and global attention, native interleaved reasoning, and support for a 256K context window, making it highly capable for local execution and complex software engineering workflows.

Key Insights & Metrics

Pricing
Free and open-weight model
Cost structure
Version
2.1
Current release version
Hardware
Requires approximately 36 GB of RAM to run effectively on local machines (e.g., Mac)
Compute requirements
Category
Open Source
Licensing model
Region
United States
Primary region

Key Features

  • 33B total parameter MoE model with 3B activated parameters
  • 256K-token context window with native interleaved reasoning support
  • Mixed sliding-window (512) and global attention (3:1 ratio)
  • FP8 KV cache quantization for reduced memory usage
  • Optimized for local deployment via Ollama, vLLM, and llama.cpp
  • Supports speculative decoding with the DFlash draft model

Discussion

0
Upvotes
0
Downvotes
0 reviews

Sign in to leave a review

Reviews

No reviews yet. Be the first to review!