Targon
Decentralized AI inference on Bittensor with confidential computing and an OpenAI API.
About
Targon, operated by Manifold Labs as Subnet 4 on the Bittensor network, is a decentralized AI inference platform where a marketplace of miners runs inference and validators verify responses. Developers get OpenAI-compatible chat, completions, image, and search endpoints over confidential, decentralized compute.
Its differentiator is confidential compute: the Targon Virtual Machine (TVM) combines hardware-backed attestation with NVIDIA confidential computing so workloads stay encrypted even on untrusted hardware. The network has aggregated around 1,600 H200s and processes over 20 billion tokens per day, with the team reporting roughly 10% savings versus traditional inference bills.
What it can do
- OpenAI-compatible API over a decentralized miner marketplace
- Confidential compute via the Targon Virtual Machine
- GPU rentals and confidential VMs behind a REST API
- MCP servers available for agent-driven provisioning
Setup
Sign up at targon.com for an API key, then point any OpenAI-compatible SDK at https://api.targon.com/v1.
Prerequisites
A Targon account with prepaid credits for the inference marketplace.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 106; Trending score 84/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.