Chutes (Bittensor subnet) Review

// Review

by the bestcloudmining editorial desk · review version 2 · prose updated 2026-09-24 · facts as of 2026-09-24 · how we verify

Overview

Chutes is a decentralised AI inference provider running on Bittensor subnet 64 (SN64), offering serverless GPU compute via an OpenAI-compatible API. It targets developers needing secure, attestable TEE-based inference with transparent per-token pricing. While it’s a live, verifiable service with real usage, capacity has historically been oversubscribed, and there’s no uptime SLA - so reliability isn’t guaranteed.

What you can rent

You can rent AI inference compute on Chutes, either through public models billed per 1M tokens or private “self-serve” chutes using RTX Pro 6000 96GB GPUs billed per second. The public API offers ~14 TEE-secured models like Mistral-Nemo and Qwen3.8, while private chutes give dedicated TEE-backed GPU access. You interact via API, CLI, or SDK. You do not own hardware, tokens, or models - only access to inference capacity.

Contract & terms

Chutes offers flexible, no-term contracts: usage is billed in real time, with no minimum commitment. There’s no maintenance fee, but private chutes have a one-time deploy fee of $5.40. Payouts aren’t applicable - this is a compute rental, not a mining service. Refunds: crypto payments are final; Stripe payments may be refunded only at Chutes’ discretion. Minimum purchase is $0.

Proof & verifiability

Yes, Chutes provides verifiable TEE (Trusted Execution Environment) compute with attestation, meaning you can cryptographically verify that models run in secure hardware enclaves. Optional end-to-end encryption ensures even Chutes can’t read your data. The public model list is accessible without auth, and usage occurs on a public Bittensor subnet - both independently checkable.

Reliability & history

Chutes is operated by Chutes Global Corp, a legally registered entity in Saint Kitts and Nevis, with published Terms and a live status page. It’s well-regarded in the Bittensor community as a working example of decentralised inference. Reddit threads from mid-2026 report past oversubscription issues that later improved. No red flags are documented, and external sources confirm its model catalog and token listing (SN64).

Pricing

Public inference starts at $0.0245 per 1M input tokens (e.g., Mistral-Nemo), rising to $3.00/$15.00 for models like Kimi K3. Private chutes use RTX Pro 6000 GPUs at $1.80/hour, billed per second, plus a $5.40 deploy fee. You can pay in USD via Stripe or in TAO. Minimum spend is $0. Pricing is public and granular, but crypto payments are final - no refunds.

Verdict

Chutes is a legitimate, verifiable option for developers wanting decentralised, TEE-secured AI inference with transparent pricing. It’s ideal if you value attestation and open access over guaranteed performance. For production-critical workloads, test thoroughly first - capacity has fluctuated, and there’s no SLA.

// Chutes (Bittensor subnet) FAQ

Is Chutes (Bittensor subnet) legit?

Based on our checks as of 2026-09-24: Chutes (Bittensor subnet) scores 6.7/10 in our weighted review. We logged no red flags. User reports and operator background are documented with dated sources in the meta-review below.

What do I actually get when I use Chutes?

You get API access to AI inference on TEE-secured GPUs, either shared (public models) or dedicated (private chutes). You don’t own hardware, tokens, or models - just compute time.

What are the total costs and fees?

Public inference is billed per 1M tokens (from $0.0245 in). Private chutes cost $1.80/hour plus a $5.40 one-time deploy fee. No maintenance fees, but crypto payments are non-refundable.

Is there a minimum spend or trial option?

No minimum - starting at $0. You can begin with small usage. There’s no formal trial, but low barriers allow testing at minimal cost.

How do payouts work?

Payouts don’t apply - this isn’t a mining or earnings service. You pay for compute used, billed in real time or per token, depending on the model.

What happens if Chutes shuts down?

If the service stops, your access ends immediately. Private chutes would terminate, and any prepaid TAO is non-refundable. No recourse is guaranteed under the Terms.

How does Chutes compare to buying hardware or coins directly?

Buying hardware gives full control but high upfront cost. Chutes offers low-entry, secure inference without ownership. It’s not an investment - compare based on compute needs, not returns.