Liquid Inference
Product Hunt Brett Harrison ● Covered by 2 sources
Liquid Inference launched an LLM router that sends each prompt to competing providers. It works with coding tools and tries to cut inference costs on the fly.
Based on reporting by Product Hunt, Brett Harrison — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Liquid Inference is launching as an LLM router, putting multiple providers in play for every prompt instead of sending the work down a single path. The pitch is simple: let providers compete in real time and push the cost down for each request.
The service says it works with a long list of agentic coding tools, including Claude Code, Codex, OpenCode, Cursor, Pi, and Cline. It also claims full multimodal support, so text isn’t the only thing it’s built to handle.
Under the hood, Liquid Inference says it can route across hundreds of open- and closed-weight models. Users can set routing-rule presets themselves or hand it over to its Auto routing algorithms. Either way, the idea is to choose the model path dynamically rather than manually picking one for every job.
There’s also a pricing twist. The platform says providers compete dynamically on price for each prompt, with the goal of offering the lowest marginal cost. New users can sign up for free with email, and the first 500 get $20 of free inference. It also offers referral rewards in the form of free inference: 20% of referred fees, and 10% for second-level referrals.
My take — AI-written commentary, not fact-checked reporting
This is the sort of product that makes sense if model sprawl is the new normal, which it plainly is. Routing is becoming the adult in the room while everyone else keeps arguing about whose model is the prettiest. Also, “providers compete for every prompt” sounds suspiciously like the internet finally learning how to bargain hunt.
Read more about this at: Product Hunt