Poolside Malibu
Frontier code foundation model for enterprises — trained with RL from code execution feedback, deployed inside your own cloud.
About
Malibu is poolside's foundation model for complex software-engineering work (its sibling Point targets low-latency completion). Poolside's distinctive training technique is RLCEF — reinforcement learning from code execution feedback — running up to 10,000 code executions per minute on AWS infrastructure to generate synthetic training data from realistic coding tasks on real-world codebases, without training on customer data. Under a multi-year AWS agreement announced at re:Invent, Malibu runs on EC2 and was slated for Amazon Bedrock, with AWS Trainium chips powering inference. The model is fine-tuned on each customer's own codebase and deployed inside their environment (VPC/on-prem), which is the whole enterprise pitch: frontier coding capability without data leaving the firewall. Poolside was valued at $3B in its October funding round (company-reported via PR). No public benchmark figures in the sources read.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
No public benchmark figures in sources read
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: AWS strategic agreement verified via PR Newswire and SiliconANGLE — Malibu + Point models, EC2/Bedrock, Trainium inference; RLCEF training technique (10k code executions/min) from SiliconANGLE; $3B valuation company-reported via PR; No public benchmark figures; enterprise-only access limits independent verification.