Liquid Inference
Architect Financial Technologies launched Liquid Inference, an LLM router that auctions each request across competing providers and charges the lowest offer that meets the buyer's rules.
Providers post offers to serve specific models, and every request is auctioned among those quoting the named model. The maximum price is locked before generation and billing covers metered usage only. It accepts OpenAI and Anthropic style API calls, and buyers can set cost caps, time to first token limits, minimum throughput, regions and zero data retention rules.
- Date
- Thursday 8 October 2026
- Lab
- Architect
- Kind
- product
- Access
- closed API
- Price
- Lowest qualifying provider offer per request; platform fee not disclosed
Figures
| Measure | Value | Measured by |
|---|---|---|
| Free inference for first users | $20 for the first 500 users per Architect's Harrison, as reported by MarkTechPost | company |
| Referral reward | 20% of referred fees as free inference, plus 10% on second-level referrals | company |
Architect also runs the AX perpetual futures exchange and plans GPU compute futures; the provider list, fees and latency data are not yet public.
Sources
This record was checked against its sources on 8 October 2026. How we check