Elyra — Private inference. Attestation on-chain.
elyra Elyra ELYRA Open app

Private inference gateway · Elysium

Private inference.
Attestation on-chain.

Prompts run inside hardware-attested enclaves the operator can't read. Every response comes back with a signed receipt, anchored on Elysium.

base_url="https://api.elyra.technology/v1"
Start building → How it works

OpenAI-compatible · receipt on every response · anchored every 10 min

RC_8F2C…A6A3 / CIPHERTEXT MOVE TO DECRYPT

01 — How it works

Five hops. No trusted middleman.

From agent to chain, every hop proves something you can check yourself.

{{ h.num }}

{{ h.name }}

{{ h.tech }}

{{ h.what }}

PROVES {{ h.proof }}

02 — Models

Open weights, list prices.

Every endpoint runs on confidential hardware, billed to the micro-dollar on the receipt.

{{ count }}
ModelSupplierInputOutputBlended ↓Receipt also proves
{{ m.id }} {{ m.supplier }} {{ m.inp }} {{ m.out }} {{ m.blend }} {{ m.proves }}

Per million tokens. Blended weights input and output 3 to 1. Prices are the suppliers' own list prices.

03 — Developers

Keep your OpenAI client.

Change the base URL. Every response now carries a receipt.

A
OpenAI-compatibleChat completions and streaming, same SDKs.
B
A receipt per responseNamed in the x-receipt-id header.
C
Docs for agentsllms.txt lets your agent read them itself.
{{ code }}

x-receipt-id: rc_8f2c… keep every receipt, verify any later

04 — Confidential hardware

Rooted in silicon.

Inference runs on hardware rooted in Intel TDX and NVIDIA Confidential Computing.

PhalaGPU TEE models on Intel TDX and NVIDIA CC
NEAR AIResponses signed in the model enclave
RedPillRouting to Phala GPU TEEs
ChutesOn Bittensor, each machine attested