Poolside
Software developers and engineering teams seeking AI-powered coding assistance and agentic coding models.
Overview
Poolside is a foundation model company focused on building open-weight AI models for software coding and agentic applications. The company develops and trains large language models from scratch, with a stated mission to bring intelligence to where work gets done and to drive abundance for humanity. Poolside emphasizes working in the open, publishing research on model training, evaluation, and iteration, and offers its models through various deployment options including on-device, desktop, CLI, and cloud platforms. Founded as an American AI company, Poolside has raised significant funding, including a $500 million Series B round led by Bain Capital Ventures, valuing the company at $3 billion. The company has also moved its headquarters to Paris, France, tapping into the European talent pool. Poolside's product lineup includes the Laguna family of models, with Laguna S 2.1 (118B params) and Laguna XS 2.1 (33B params) as its latest releases, designed for agentic coding tasks with a focus on quality, speed, and efficiency.
What it does
Poolside builds and trains open-weight foundation models specifically for agentic coding. The models are designed to run in various environments, from on-device to cloud, and are optimized for reasoning, speed, and efficiency. The company provides a platform for deploying these models, offering complete control of model weights and predictable subscription plans. Poolside also develops a runtime for training and operating agents, and offers product experiences in research preview.
Key features
- Open-weight foundation models for coding
- Agentic coding capabilities
- Multiple model sizes (Laguna S 2.1: 118B params, 8B active, 1M context; Laguna XS 2.1: 33B params, 3B active, 256K context)
- On-device and cloud deployment options
- Desktop and CLI interfaces
- Research publications on model training and evaluation
- Government-focused offerings
Use cases
- AI-assisted software development
- Automated code generation and completion
- Code review and refactoring
- Planning and executing software engineering tasks
- Running coding agents on-device or in the cloud
Pricing
Pricing is not publicly listed; custom pricing based on usage, with predictable subscription plans for deployments.
Pricing is gathered from the company's public pages and may change. Check the vendor's site before buying.
Pros and cons
Strengths
- Open-weight models allow for customization and control
- Multiple model sizes to fit different computational needs
- Strong focus on research and transparency
- Flexible deployment options including on-device and cloud
- Backed by significant funding and major investors
Limitations
- Pricing not publicly disclosed; custom pricing based on usage
- Primarily focused on coding, limiting applicability to other domains
- Early-stage product with research previews for some features
What sets it apart
- Open-weight models with a focus on agentic coding
- Models trained from scratch, not fine-tuned from existing open models
- Strong emphasis on research and publishing technical details
- On-device capability with small model sizes
Ecosystem
Integrations
Frequently asked questions
What models does Poolside offer?
Poolside offers the Laguna family of models, including Laguna S 2.1 (118B params, 8B active, 1M context) and Laguna XS 2.1 (33B params, 3B active, 256K context).
Can I run Poolside models on my own hardware?
Yes, the Laguna XS 2.1 model is small enough to run on-device, and Poolside provides desktop and CLI interfaces for local use.
How can I access Poolside models?
You can get started via desktop, CLI, or through platforms like OpenRouter and Vercel AI Gateway.
Is Poolside's pricing public?
No, pricing is not publicly listed. Poolside offers custom pricing based on usage, with predictable subscription plans for deployments.
What is Poolside's approach to AI safety?
Poolside publishes research on model training and evaluation, including discussions on reward hacking and strategies to address it.
Sources
This profile was compiled from the company's own pages and public web research.
Last researched August 11, 2026.
Manage this listing
Own this company? Update your listing information.