Liquid Inference
An LLM inference routing product that uses a real-time reverse auction mechanism, allowing multiple inference providers to bid on each prompt and automatically routing requests to the lowest-cost provider. It supports both rule-based preset routing and automatic routing modes.
Core Facts
Timeline (last 90 days)
Cline CLI v3.0.74's model catalog adds two providers, Liquid Inference and Munito AI, and removes CrofAI, iFlow and Llama.
The model catalog decreased from 7,041 to 6,457, adding three providers—Liquid Inference, Munito AI and Munito AI (EU)—and removing CrofAI, iFlow and Llama.
Liquid Inference supports mainstream agent coding tools including Claude Code, Codex, OpenCode, Cursor, Pi, and Cline.
Liquid Inference claims to support full multimodal capabilities and has integrated hundreds of open-weight and closed-weight models.
Liquid Inference offers two routing modes: user-defined routing rule presets and the official Auto automatic routing algorithm.
The first 500 users of Liquid Inference can receive $20 in free inference credits.
Liquid Inference's referral mechanism returns 20% of fees generated by referred users as free inference credits, with second-level referrals returning 10%.
Liquid Inference is an LLM router product launched on Product Hunt, whose core mechanism allows multiple inference providers to dynamically bid when a user sends a prompt, with the provider offering the lowest marginal cost handling the request.
On its launch day, Liquid Inference received 74 upvotes and 4 comments on Product Hunt, ranking 13th for the day.
Bidding-based routing faces a model consistency issue: when the same request is executed by different providers, output quality, style, and stability may vary, posing a challenge for production scenarios that require determinism.
4 more timeline events
All Facts (11)
Cline CLI v3.0.74's model catalog adds two providers, Liquid Inference and Munito AI, and removes CrofAI, iFlow and Llama.
75%UnverifiedThe model catalog decreased from 7,041 to 6,457, adding three providers—Liquid Inference, Munito AI and Munito AI (EU)—and removing CrofAI, iFlow and Llama.
50%UnverifiedLiquid Inference supports mainstream agent coding tools including Claude Code, Codex, OpenCode, Cursor, Pi, and Cline.
50%UnverifiedLiquid Inference claims to support full multimodal capabilities and has integrated hundreds of open-weight and closed-weight models.
50%UnverifiedLiquid Inference offers two routing modes: user-defined routing rule presets and the official Auto automatic routing algorithm.
50%UnverifiedLiquid Inference is an LLM router product launched on Product Hunt, whose core mechanism allows multiple inference providers to dynamically bid when a user sends a prompt, with the provider offering the lowest marginal cost handling the request.
50%UnverifiedBidding-based routing faces a model consistency issue: when the same request is executed by different providers, output quality, style, and stability may vary, posing a challenge for production scenarios that require determinism.
50%UnverifiedThe introduction of a bidding decision layer may increase response latency, representing an engineering trade-off in bidding-based routing implementations.
50%UnverifiedThe value of a bidding-based routing system is highly dependent on the scale of the supply-side ecosystem — the more providers participating in bidding and the greater the price spread, the more apparent the system's value.
50%UnverifiedThe Maker of Liquid Inference is Brett Harrison.
50%UnverifiedLiquid Inference adopts a reverse auction mechanism — in contrast to programmatic advertising's real-time bidding that seeks the highest bidder, inference bidding seeks the lowest-cost supplier.
50%