DoubleAI launches Maxwell, a model that optimizes code at runtime

By Splitfeed EditorialLast updated Oct 19, 2026, 19:00 UTC

On October 19, 2025, Gal, a founder at DoubleAI, announced Maxwell on Hacker News, describing it as a Code Execution Model. The system positions itself as a layer between application code and hardware, optimizing execution speed in real time without altering source code. Maxwell does this by observing programs from within, drawing on traces, profiling data, and live input characteristics to adapt to the hardware environment and whatever other processes are running at the same time. The announcement included no performance benchmarks, pricing information, or release dates.

The core claim is technically ambitious. Runtime optimization is not new, compilers and just-in-time systems have approached this problem for decades, but Maxwell's framing as a frontier model that continuously adapts during execution, rather than ahead of it, positions the approach as meaningfully different. The absence of benchmark data makes independent assessment impossible at this stage, which is a significant gap given that performance claims are the entire basis of the product's value. The explicit statement that no humans are in the loop is notable because it describes an AI system making continuous, real-time decisions about how software executes, with no point of human review.

None of the three camps have published reactions to this announcement yet. The Pro-AI camp would typically welcome a system like Maxwell as an example of AI delivering concrete, measurable productivity gains for developers without requiring workflow changes. The Anti-AI camp would typically raise concerns about the absence of human oversight in a system that intervenes in live code execution, and would likely press hard on the lack of benchmarks as a reason for skepticism. The Middle Ground camp would typically acknowledge the technical interest while calling for rigorous, independent performance validation before treating the claims as established.

The most important thing to watch for is the release of concrete benchmark data, ideally from parties outside DoubleAI, showing what speed improvements Maxwell actually delivers across different hardware environments and workload types. Pricing and availability details would also clarify whether this is a research announcement or a product approaching general use.