Poolside Overview
This Poolside review covers one of the more unusual bets in AI coding for 2026: a company building foundation models trained exclusively on code, rather than bolting code features onto a general chatbot. Poolside’s Malibu and Point models power enterprise software development with self-hostable open weights — and an August 2026 licensing deal with NVIDIA worth a reported $6 billion put the technology on every CTO’s radar.
What is Poolside?
Poolside is an AI company developing coding-specialized foundation models from scratch. Its Malibu model handles multi-file code generation, refactoring, and test and documentation generation, while Point is optimized for sub-200-millisecond IDE code completion. Because the models are purpose-trained on code rather than adapted from chat models, Poolside argues they understand software engineering workflows more natively.
The models are trained with RLCEF — reinforcement learning from code execution feedback — meaning the training loop rewards code that actually runs correctly, not just code that looks plausible. Deployments support up to 1M-token context windows, and an Amazon Bedrock partnership brings the models to AWS customers.
Crucially for regulated industries, Poolside releases open-weight versions (such as Laguna S 2.1) that enterprises can download, self-host, and fine-tune on their own codebases — including quantized GGUF builds for local use. That self-hosting story is the core enterprise pitch: your code never has to leave your infrastructure.
Poolside Pricing in 2026
There is no public per-seat pricing page; enterprise deals are custom quotes. The accessible entry points are the free open-weight model downloads for self-hosting and token-based API access via OpenRouter at roughly $0.10 per million input tokens and $0.20 per million output tokens — an order of magnitude cheaper than frontier-lab coding APIs. Individual developers should treat Poolside as an infrastructure option, not a consumer product: there is no Copilot-style $10-a-month seat to buy.
Who Should Use Poolside?
Enterprise engineering organizations with strict data-residency requirements are the target: banks, governments, and large tech firms that want AI coding assistance without sending proprietary code to a third-party API. Teams already on AWS can access the models through Bedrock, and ML-ops-savvy orgs can self-host the open weights.
It is not for indie developers or small teams who want a plug-in assistant today — there is no simple product for them. Anyone choosing Poolside should also weigh the company’s turbulent 2026: a reported $2 billion Series C collapsed before the NVIDIA deal steadied the ship.
Our Verdict on Poolside
Poolside’s technology is legitimately impressive — code-native models with RLCEF training and genuinely cheap inference. But in 2026 it remains an enterprise infrastructure play, not a developer product. The NVIDIA deal validates the tech; the missing piece is a product mere mortals can buy.
The bottom line of this Poolside review: exciting models, enterprise-only access. For a coding assistant you can use today, compare Cursor or Aider instead. Browse more in our AI Developer Tools category.
Key Features
- Coding-specialized foundation models (Malibu, Point)
- RLCEF training — models learn from code that actually executes
- Open-weight releases free for download and self-hosting
- Up to 1M-token context windows for large codebases
- Quantized GGUF builds for local deployment
- Amazon Bedrock partnership for AWS-native access
- Fine-tuning on enterprise codebases
Poolside Pricing
| Plan | Price |
|---|---|
| Open weights | $0 |
| Token API | ~$0.10/1M input |
| Enterprise | Custom |
Pricing checked on October 5, 2026 — always confirm on the official site.
Poolside Pros & Cons
✓ Pros
- Models purpose-trained on code, not adapted from chat
- Self-hostable open weights suit regulated enterprises
- Token economics an order of magnitude cheaper than frontier APIs
- RLCEF training rewards code that actually runs
- NVIDIA's $6B licensing deal validates the technology
✕ Cons
- No public per-seat pricing or self-serve product for individual developers
- Company's financial path looked shaky in 2026 after a collapsed funding round
- Enterprise-only go-to-market leaves small teams and indie devs underserved
- Competes directly with frontier labs like Anthropic and OpenAI on capability
Poolside FAQs
How much does Poolside cost in 2026?
Is Poolside free to use?
What are Poolside's Malibu and Point models?
Is Poolside still in business?
What are the best Poolside alternatives?
Best Poolside Alternatives
Continue
★★★★☆Continue reviewed for 2026 — open-source AI coding assistant for VS Code and JetBrains, any-model support, local privacy. Free core, $10/dev team.
Pinecone
★★★★☆Pinecone reviewed for 2026 — the serverless vector database for RAG and AI apps. Pricing from free, real pros and cons, and when to choose it.
OpenHands
★★★★☆OpenHands reviewed for 2026 — the open-source AI coding agent with 89k+ stars. Free self-hosting, model-agnostic design, honest pros, cons, and verdict.
Builder.io
★★★★☆Builder.io reviewed for 2026 — Fusion's AI visual IDE writes framework-native code into your own repo. Features, pricing from $30/user/mo, honest pros and cons.