category.decentralised_compute.label

category.decentralised_compute.promise

Starting path

The advertised fleet is a claim. The API returns a number.

Decentralised Compute evaluated across real cost per 1m identical calls, archive, debug and trace access (and what it really costs), published throughput and self-serve ceiling, and documented incident record and chain coverage.

category.decentralised_compute.promise Reviewed on a fetchable price per GPU-hour, supply verified against the network's own API, published utilisation, and recourse when a provider vanishes mid-job.

Start with the smart category assistant or jump straight into quick questions. Both paths narrow to the same recommendation standard.

How we review

Recommendation first, comparison only if needed.

Reviewed and rechecked on the current cycle.

Commercial relationships disclosed before clickout.

Audit

Infrastructure

Methodology

2026.04.0

published 2026-04-27

Providers tracked

570+

across 62 categories

Last verified

2026-08-05

editorial freshness

Editorial board

4

audit-log live

Decision guide

What matters most before choosing in this category

The best compute network depends on your call mix and your throughput ceiling — but headline plan prices in this category are not comparable at all. Every vendor bills in its own unit, and the same eth_call is 1 unit at one provider and 200 at another, so a plan that looks cheaper per month can be many times dearer for identical work. The scored figure is the cost of one fixed workload derived from each vendor's own published unit table, and incident history is weighted by blast radius rather than count.

Key question

How long does a single job run, and can it restart if a provider drops out?

Key question

Do you need a specific GPU class, and is it actually available today?

Key question

What recourse do you have if a supplier vanishes mid-job?

Current editor lead

Akash Network

Cosmos-based compute marketplace; live GPU prices and inventory via public API

Data checked Aug 2026

Cosmos-based compute marketplace; live GPU prices and inventory via public API. Strongest on fetchable price per gpu-hour (9/10): The only network in the category with a machine-readable public price feed: console-api.akash.network/v1/gpu-prices returns min/max/avg/weighted-average/median USD per hour for 24 GPU models, no key or login. H100 SXM5 resolves to min $2.01 / med $2.56 / max $3.16 per hour. Weakest on recourse when a provider vanishes mid-job (3/10): Akash's own core-concepts docs define a provider-terminated lease as 'Provider closed it. Permanent end state.' and the prescribed remedy for provider downtime is 'If persistent, close and redeploy to different provider.' No refund, credit, or notice period. 'uptime-guarantee: 99.9%' appears only as an optional… Published price: Live from Akash's own console API (fetched 5 Aug 2026): h100 80Gi SXM5 — min $2.01, max $3.16, avg $2.57, med $2.56/hr, 4 providers; a100 80Gi SXM4 — min $1.13, max $1.83, avg $1.55/hr; h200 141Gi SXM5 — $4.37/hr flat, 2 providers; rtx4090 24Gi —…

Best for

Fetchable Price Per GPU-Hour — 9/10Published Utilisation and Live Availability — 9/10

Main tradeoff

The entire network is 445 GPUs, and H100 capacity is 71 cards held by four providers. One operator withdrawing is a double-digit-percent capacity event, and there is no reservation or queue mechanism to protect an in-flight workload. Price transparency here is genuinely best-in-class; scale is not.

Verify before signup

Live from Akash's own console API (fetched 5 Aug 2026): h100 80Gi SXM5 — min $2.01, max $3.16, avg $2.57, med $2.56/hr, 4 providers; a100 80Gi SXM4 — min $1.13, max $1.83, avg $1.55/hr; h200 141Gi SXM5 — $4.37/hr flat, 2 providers; rtx4090 24Gi — $0.28–$0.38/hr; rtx5090 32Gi — $0.59–$0.71/hr; gtx1050 2Gi — $0.06/hr. Prices returned in both USD and uakt across 24 NVIDIA models.

Weighted criteria

Fetchable Price Per GPU-Hour30%

Verified vs Self-Reported Supply30%

Published Utilisation and Live Availability20%

Recourse When a Provider Vanishes Mid-Job20%

Leading options

Shared shortlist for this category

These providers are pulled from the same category comparison catalog used in validation, so the category page, comparison page, and provider reviews stay aligned.

Browse this network

Methodology

How this category is reviewed

Reviewed on a fetchable price per GPU-hour, supply verified against the network's own API, published utilisation, and recourse when a provider vanishes mid-job.

Reviewed on: Fetchable Price Per GPU-Hour, Verified vs Self-Reported Supply, Published Utilisation and Live Availability, Recourse When a Provider Vanishes Mid-Job.

This page is a maintained category surface, not a static marketing block. Review freshness, provider positioning, and recommendation logic should stay consistent with quiz and provider pages.

Frequently asked

Questions people ask before choosing compute

Why are advertised node counts not scored?

Because they are self-reported, and this category has a documented history of inflated supply. What is scored is the number the network's own API returns today, against the price you can fetch for it.

Is idle capacity the same as available capacity?

No. A large fleet sitting mostly idle can still be unavailable in the GPU class you need, which is why published utilisation is scored separately from raw supply.

What happens when a permissionless supplier disappears?

That is the criterion. A permissionless marketplace has no SLA behind an individual supplier, so what matters is whether the network re-schedules your job, refunds you, or leaves you to notice.

REVIEWEDApr 2026METHOD4 criteriaCATEGORYdecentralised_compute

Not financial advice · For informational purposes only · Always do your own research

Octopus · The AI CFO that pays for itself

Audit your entire crypto stack — free.

Score concentration, fees, security, and tax complexity across exchanges, wallets, staking, and DeFi.

Not financial advice · For informational purposes only · Always do your own research