Chutes Integration

Chutes Integration

Chutes SN64 is the immediate decentralized inference provider target for XEL.

Role In XEL

When a Living Character has an encrypted persona artifact and a user sends a chat request, the XEL runtime can select Chutes from provider_policy and send an OpenAI-compatible chat completion request.

Flow:

1. XEL checks profile visibility, password/payment access, spend limits, and runtime provider policy.
2. Runtime obtains the allowed persona prompt/context through the configured access policy.
3. Runtime sends an OpenAI-compatible request to Chutes.
4. Chutes generates the response and returns usage/model metadata.
5. XEL records usage, payment receipt, provider ID, model ID, and optional proof refs.

Default Model Candidate

Live model target verified from https://llm.chutes.ai/v1/models on 2026-07-06:

Qwen/Qwen3-32B-TEE

The model is selected as the first low-cost TEE-marked Chutes target for MVP testing. Owner policy can later select a larger or more roleplay-specific model from the live Chutes catalog.

XEL may support adult-capable, human-like characters through owner-selected open-weight models. That does not remove XEL policy boundaries: the runtime must still block minors, non-consensual sexual content, coercion, exploitation, illegal sexual content, and privacy abuse before or around the model call.

Local Example

The generic runtime router can call Chutes from examples/provider-registry.xel.json and examples/provider-policy.chutes.json.

For the local simulator:

CHUTES_API_KEY=cpk_... node runtime/simulate-chat-turn.mjs --funded --provider chutes

The adapter lives at:

runtime/providers/chutes.mjs

Payment Notes

Chutes exposes an OpenAI-compatible inference base URL:

https://llm.chutes.ai/v1

The Chutes inference API currently uses Bearer API-key auth for direct server-to-server calls. XEL therefore treats x402 as a proxy or facilitator rail for Chutes until Chutes exposes native x402 on the selected inference endpoint.

The first deployment needs one of:

• a Chutes API key funded through the XEL treasury;
• an XEL/GEN payment facilitator that accepts x402 and forwards to Chutes with a funded API key;
• a character-funded operating balance on a provider-supported x402 proxy rail.

Remaining environment and secret inputs for a real paid Chutes call:

Name	Secret	Purpose
CHUTES_API_KEY	Yes	Bearer token for the Chutes OpenAI-compatible inference API.
CHUTES_BASE_URL	No	Override for https://llm.chutes.ai/v1.
XEL_X402_FACILITATOR_URL	No	XEL or third-party facilitator endpoint that can verify/settle x402 payment payloads before forwarding to Chutes.
XEL_X402_FACILITATOR_TOKEN	Yes	Optional facilitator API credential when the facilitator is not public verifier-only.
XEL_RUNTIME_PAYMENT_WALLET_ID	No	Wallet descriptor ID from payment-capability.v1, normally the character’s inference_runtime wallet.
XEL_RUNTIME_PAYMENT_SIGNER_REF	Yes	Reference to testnet signer/key material held in Infisical or an HSM/KMS; do not store raw keys in repo files.

Do not hard-code Chutes as the only inference provider. It is the immediate decentralized provider target, with Verathos SN96, Targon SN4, Phala, Marlin/Oyster, Akash-hosted open models, and GEN runtime as alternate/fallback provider classes.