AI infrastructure / 10 companies
Inference. Serving models — making each answer cheaper, faster, or runnable on more hardware.
Companies
Newest on the page first. Sorted by date only.- 9 Oct 2026Inference.netHosts custom and open models, and traces and tunes the agents that call them.Following
- 9 Oct 2026MorphFast edit, search and context-compaction models for coding agents.Following
- 9 Oct 2026RelaceFast small models and tools for coding agents, plus hosted open models.Following
- 9 Oct 2026TensormeshA shared KV cache that reuses work a model already did, hosted or on your own cluster.Following
- 15 Sept 2026CallosumSplits one AI job across whichever models and chips run each part cheapest.Following
- 15 Sept 2026DoublewordHosted open-model inference at realtime, async and overnight-batch prices.Following
- 15 Sept 2026FlexAIOne API key for open models that grows from per-token to reserved hardware.Following
- 15 Sept 2026Pruna AICompresses models so the same output costs less GPU time.Following