Codeflash

Codeflash is an AI performance engineering product that finds and ships optimizations to existing code. Its agent explores a codebase in an isolated sandbox, proposes faster implementations, benchmarks each change, verifies correctness against existing tests plus automatically generated regression tests, and delivers the result as a reviewable pull request with benchmark numbers and rationale attached. Codeflash currently optimizes Python only. The company now leads with a managed engagement aimed at companies that want to cut infrastructure cost, latency, memory use or GPU load. An engagement starts by choosing an objective and reproducing a baseline on a representative workload; the agent then optimizes autonomously, the company's own performance engineers review every change before it reaches the customer, and the agent is then set to optimize each new pull request. The vendor states that customer code is never used to train models. The self serve product runs against GitHub projects, with tracing, developer code quality metrics and a user dashboard on paid tiers. The pricing page lists self serve plans. Free costs $0 with 25 function optimization credits per month, community support, public GitHub projects only, local tracing only, and AI training permitted on the code. Pro costs $20 per user per month after a 14 day free trial, with 500 optimization credits per user per month, a zero data retention policy, private GitHub projects, priority support and analytics. Enterprise is custom priced and adds unlimited credits, admin analytics, on premises deployment, 24/7 support and custom SLAs. The managed performance engagement is sold through a call, and its pricing is not published.