# Confidential Inference Directory

> Independent directory and zero-trust attestation monitor for confidential-inference (TEE) AI providers: hardware attestation, workload identity, source provenance, request routing and pricing compared per provider and per model. Anything the stored evidence cannot support is shown as unknown; provider claims are labelled as claims.

## Pages

- [Providers](https://confidentialinference.net/providers): Every catalog provider in one dense table — declared TEE hardware, trust chain, prices, attestation monitor results.
- [Models](https://confidentialinference.net/models): Exact per-model offerings with current price sources and attestation state.
- [Route Explorer](https://confidentialinference.net/routes): Who sells a model, who routes it, who operates its confidential workload — per-route evidence strength.
- [Attestation Monitor](https://confidentialinference.net/attestation): Current zero-trust verification results per provider and check.
- [Methodology](https://confidentialinference.net/methodology): How checks run, the seven evidence gates, status meanings, what a TEE does not prove, and known physical attacks on server TEEs.
- [FAQ](https://confidentialinference.net/faq): Plain-language answers about TEEs, attestation, E2EE, routers and verification results.
- [Weekly](https://confidentialinference.net/weekly): Weekly attestation-coverage digest.
- [Feeds](https://confidentialinference.net/feeds): RSS/JSON alert feeds for verdict changes.
- [TLS vs E2EE architecture](https://confidentialinference.net/architecture/tls-vs-e2ee): Channel topologies and what each proves.

## APIs

- [Providers JSON](https://confidentialinference.net/api/providers.json): Full catalog snapshot (providers, models, pricing, freshness).
- [Routes JSON](https://confidentialinference.net/api/v1/routes.json): Observed route evidence per check.
- [Changes JSON](https://confidentialinference.net/api/v1/changes.json): Durable verdict-change log with cursor pagination.
- [Attestation history JSON](https://confidentialinference.net/api/v1/attestation/history.json): Per-provider attestation run history.
- [Per-provider JSON](https://confidentialinference.net/api/providers/{slug}.json): One provider's catalog entry (replace {slug}).

## Providers

- [Tinfoil](https://confidentialinference.net/providers/tinfoil): Verifiable private inference on AMD SEV-SNP and Intel TDX with confidential NVIDIA GPUs. 18 models.
- [NanoGPT](https://confidentialinference.net/providers/nanogpt): Pay-per-prompt confidential inference — TEE-backed models with ECDSA per-request attestation. 20 models.
- [Confidential AI](https://confidentialinference.net/providers/confidentialai): Confidential inference on Intel TDX with a nonce-bound, TLS-bound front-door attestation. 1 models.
- [Prem](https://confidentialinference.net/providers/prem): OpenAI-compatible end-to-end encrypted confidential inference on dedicated GPUs. 3 models.
- [Enclave42](https://confidentialinference.net/providers/enclave42): OpenAI-compatible TEE inference with a quote-bound E2EE client and signed security receipts. 11 models.
- [Cohere Model Vault (Encrypted)](https://confidentialinference.net/providers/cohere): Dedicated single-tenant confidential deployments priced per GPU-hour, reached through a client-side OHTTP proxy. 2 models.
- [Redpill](https://confidentialinference.net/providers/redpill): OpenAI-compatible confidential inference across attested GPU TEE routes. 2 models.
- [Chutes](https://confidentialinference.net/providers/chutes): Decentralized confidential inference on Bittensor with TEE. 11 models.
- [NEAR AI](https://confidentialinference.net/providers/near): Private, verifiable AI — Intel TDX + NVIDIA H200 TEE with on-chain attestation. 16 models.
- [Maple](https://confidentialinference.net/providers/maple): Privacy-first inference with AWS Nitro Enclave attestation. 10 models.
- [Privatemode](https://confidentialinference.net/providers/privatemode): EU-hosted confidential inference via Cosmian VM. 7 models.
- [PPQ.AI](https://confidentialinference.net/providers/ppq): Private TEE inference via hardware-attested confidential compute. 7 models.
- [Venice.ai](https://confidentialinference.net/providers/venice): End-to-end encrypted private AI with TEE attestation. 13 models.