gpurank
esc
  • Nothing named “”. Press Enter to search the offers instead.
↑↓ to move · ↵ to open · esc to close0 entries
Marketplace · San Francisco, CA · founded 2018

Akash Network

Decentralized compute marketplace on the Akash blockchain where independent providers bid for container deployments; managed Akash Console billed by card, or self-custody wallet deployments paid in ACT/AKT.

GPU types28with a current offer
Current offers7677 readings on file · last 45 days
H100 from$2.04per GPU-hour, on-demand
Entry price$0.06cheapest GPU-hour listed
Countries2136 named regions
Trust Score v0.1
53/100
  • Company & track record100
  • Infrastructure control60
  • Pricing & billing transparency80
  • Reliability signals25
  • Security & compliance0
  • Customer signal—

Scored Sep 29, 2026 from the provider's own pages. Dimensions not yet measured are excluded from the denominator. Criteria ↓

Trust Score · 5 of 6 dimensions measured

The score, criterion by criterion

Each point comes from something the provider publishes, with the page it was read on and the date. Anything not disclosed earns zero.

Company & track record
weight 15%
100 / 100
Founding year published2018akash.network · Sep 29, 20264 / 4
Headquarters publishedOverclock Labs Inc.akash.network · Sep 29, 20263 / 3
Legal entity namedOverclock Labs Inc.akash.network · Sep 29, 20264 / 4
Funding or ownership disclosedOpen-source protocol with a native token (AKT) used for staking, governance and gas; stats page on 2026-09-29: 90.17M of 298.31M AKT bonded, inflation 4.00%akash.network · Sep 29, 20264 / 4
Infrastructure control
weight 15%
60 / 100
Capacity model publishedthird-party hostsakash.network · Sep 29, 20264 / 10
Regions published36 regions or countriesakash.network · Sep 29, 20265 / 5
Pricing & billing transparency
weight 20%
80 / 100
Public prices without an accountyes (read on the pricing page)akash.network · Sep 29, 20265 / 5
Billing granularity publishedper block (about 6 seconds, 600 blocks per hour) from an on-chain escrowakash.network · Sep 29, 20264 / 4
Egress policy publishednot disclosedakash.network · Sep 29, 20260 / 4
Stopped-instance billing publishedNo stop state: a deployment runs until closed or until escrow is depleted. Closing returns unused escrow; persistent storage is provider-local and is lost when the deployment is closed or moved to another provider.akash.network · Sep 29, 20264 / 4
Public IPv4 policy publishedextra: dedicated IPv4 via IP lease, priced by the provider; docs cite 'typically ~$5-15/month additional'. Default exposure through provider ingress hostnames and ports.akash.network · Sep 29, 20263 / 3
Reliability signals
weight 20%
25 / 100
Status pagenot disclosedakash.network · Sep 29, 20260 / 6
SLA publishednot disclosedakash.network · Sep 29, 20260 / 6
SLA with a numeric uptime commitmentnot disclosedakash.network · Sep 29, 20260 / 3
Uptime history publishedPer-provider uptime (1d/7d/30d) in the public Console providers API; network stats at stats.akash.networkakash.network · Sep 29, 20265 / 5
Observed availability (30 days)not measured yet—
Security & compliance
weight 15%
0 / 100
SOC 2not disclosedakash.network · Sep 29, 20260 / 8
ISO 27001not disclosedakash.network · Sep 29, 20260 / 4
HIPAA or GDPR DPAnot disclosedakash.network · Sep 29, 20260 / 3
Customer signal
weight 15%
not measured
Aggregated public reviewsnot collected yet—
provisional v0.1: evidence from the provider's own pages; dimensions not yet measured are excluded from the denominatorFull method →
Prices by GPU · 19 GPU types with a current offer

What Akash Network charges, against the market

Cheapest current on-demand true cost here, next to the market median for the same GPU. Negative means cheaper than the market.

GPUCheapest hereMarket medianvs marketCurrent offersModes
H100$2.04$3.17−36%1on-demandH100 prices
A100 80GB$1.07$1.99−46%1on-demandA100 80GB prices
RTX PRO 6000$1.44$2.14−33%2on-demandRTX PRO 6000 prices
RTX 5090$0.57$0.69−17%1on-demandRTX 5090 prices
T4$0.11$0.51−78%1on-demandT4 prices
RTX 4090$0.36$0.44−18%1on-demandRTX 4090 prices
Tesla V100 32GB$0.18$2.48−93%1on-demandTesla V100 32GB prices
Tesla V100 16GB$0.25$1.71−85%1on-demandTesla V100 16GB prices
RTX 3090$0.09$0.19−53%1on-demandRTX 3090 prices
RTX 4000 Ada$0.29$0.27+8%1on-demandRTX 4000 Ada prices
RTX 3060 Ti$0.14$0.07+115%1on-demandRTX 3060 Ti prices
RTX 4080$0.15$0.25−40%1on-demandRTX 4080 prices
RTX 3060$0.08$0.05+48%1on-demandRTX 3060 prices
Tesla P4$0.06$0.07−11%1on-demandTesla P4 prices
Tesla P40$0.15$0.09+63%1on-demandTesla P40 prices
A10G$0.18$1.01−82%1on-demandA10G prices
Quadro RTX 5000$0.18$0.18−0%1on-demandQuadro RTX 5000 prices
GTX 1050$0.14——1on-demandGTX 1050 prices
RTX 4000 SFF Ada$0.14——1on-demandRTX 4000 SFF Ada prices
All offers · 77 readings on file

Every Akash Network offer, with its receipt

Current offers (last 45 days) ranked by true cost, all rental modes, each with its read date. Older readings stay listed below, in grey.

GPUConfigurationModeRegionStockListTrue costRead
Tesla P4marketplace floor1× Tesla P4On-demandAkash Network providers (decentralised)In stock$0.06$0.0621:10 UTCfreshReceipt
RTX 30601× RTX 3060On-demandAkash Network providers (decentralised)Unavailable$0.08$0.0821:10 UTCfreshReceipt
RTX 3090marketplace floor1× RTX 3090On-demandAkash Network providers (decentralised)In stock$0.09$0.0921:10 UTCfreshReceipt
T4marketplace flooroutlier · suspended1× T4On-demandAkash Network providers (decentralised)In stock$0.11$0.1121:10 UTCfreshReceipt
GTX 1050marketplace floor1× GTX 1050On-demandAkash Network providers (decentralised)In stock$0.14$0.1421:10 UTCfreshReceipt
RTX 3060 Timarketplace floor1× RTX 3060 TiOn-demandAkash Network providers (decentralised)In stock$0.14$0.1421:10 UTCfreshReceipt
RTX 4000 SFF Ada1× RTX 4000 SFF AdaOn-demandAkash Network providers (decentralised)Unavailable$0.14$0.1421:10 UTCfreshReceipt
Tesla P40marketplace floor1× Tesla P40On-demandAkash Network providers (decentralised)In stock$0.15$0.1521:10 UTCfreshReceipt
RTX 40801× RTX 4080On-demandAkash Network providers (decentralised)Unavailable$0.15$0.1521:10 UTCfreshReceipt
A10Gmarketplace flooroutlier · suspended1× A10GOn-demandAkash Network providers (decentralised)In stock$0.18$0.1821:10 UTCfreshReceipt
Quadro RTX 5000marketplace floor1× Quadro RTX 5000On-demandAkash Network providers (decentralised)In stock$0.18$0.1821:10 UTCfreshReceipt
Tesla V100 32GBmarketplace flooroutlier · suspended1× Tesla V100 32GBOn-demandAkash Network providers (decentralised)In stock$0.18$0.1821:10 UTCfreshReceipt
Tesla V100 16GBmarketplace flooroutlier · suspended1× Tesla V100 16GBOn-demandAkash Network providers (decentralised)In stock$0.25$0.2521:10 UTCfreshReceipt
RTX 4000 Adamarketplace floor1× RTX 4000 AdaOn-demandAkash Network providers (decentralised)In stock$0.29$0.2921:10 UTCfreshReceipt
RTX 4090marketplace floor1× RTX 4090On-demandAkash Network providers (decentralised)In stock$0.36$0.3621:10 UTCfreshReceipt
RTX 50901× RTX 5090On-demandAkash Network providers (decentralised)Unavailable$0.57$0.5721:10 UTCfreshReceipt
A100 80GBmarketplace floor1× A100 80GBOn-demandAkash Network providers (decentralised)In stock$1.07$1.0721:10 UTCfreshReceipt
RTX PRO 6000marketplace floor1× RTX PRO 6000On-demandAkash Network providers (decentralised)In stock$1.44$1.4421:10 UTCfreshReceipt
RTX PRO 60001× RTX PRO 6000On-demandAkash Network providers (decentralised)In stock$1.52$1.5221:10 UTCfreshReceipt
H100SXMmarketplace floor1× H100 SXMOn-demandAkash Network providers (decentralised)In stock$2.04$2.0421:10 UTCfreshReceipt
20 fresh offers · prices per GPU-hour · true cost includes the add-ons the provider publishesOpen in the comparison tool →
True cost & billing · read Sep 29, 2026

What the hourly price does not say

Billing granularity, minimums, what a stopped machine costs, traffic, addresses, deposits and identity checks, as published by the provider.

Billing

Billing granularity
per block (about 6 seconds, 600 blocks per hour) from an on-chain escrow
Minimum charge
No minimum rental period published. Self-custody deployments need an escrow deposit of about 5 ACT minimum plus AKT gas (about 0.02 to 0.1 AKT per transaction); Console Air ACT mint minimum 10 ACT.
Stopped instance
No stop state: a deployment runs until closed or until escrow is depleted. Closing returns unused escrow; persistent storage is provider-local and is lost when the deployment is closed or moved to another provider.
Egress
not published
Public IPv4
extra: dedicated IPv4 via IP lease, priced by the provider; docs cite 'typically ~$5-15/month additional'. Default exposure through provider ingress hostnames and ports.
Billing model
prepaid credits
Minimum deposit
Managed Console: none published (card top-up via Stripe). Self-custody: about 5 ACT escrow per deployment.
Free credits
Docs: 1 USD trial credit, no card, 30 days, deployments limited to 24 h each, GPUs up to RTX 3090 only. Marketing pages show other figures (blog CTA 'Start with $100 Free Credits'; Dec 2024 announcement 10 USD).

Account

Payment
card · crypto (yes: self-custody path only (Console Air, CLI, SDK) with a Keplr/Cosmos wallet: escrow funded in ACT (USD-pegged compute credit minted by burning AKT); AKT accepted for escrow top-ups only when the circuit breaker is active; the API docs page still lists 'AKT or USDC' for the blockchain SDK. The managed Console no longer accepts wallet connections (card only).)
Sign-up
email (6-digit code, no password) or Google or GitHub account, card to go past the trial (managed Console), Keplr-compatible wallet for self-custody (no account)
Identity check
ID on request
Support
discord, github issues, github discussions, contact form
Enterprise support
not published
SLA
not published
Status page
not published
Published uptime
Per-provider uptime (1d/7d/30d) in the public Console providers API; network stats at stats.akash.network

Infrastructure

Regions
36 named regions in 21 countriesNew South Wales (provider IP location) (Australia) · Flanders (provider IP location) (Belgium) · Sofia-Capital (provider IP location) (Bulgaria) · Prague (provider IP location) (Czechia) · Uusimaa (provider IP location) (Finland) · Hesse (provider IP location) (Germany) · Saxony (provider IP location) (Germany) · Guatemala (provider IP location) (Guatemala) · Maharashtra (provider IP location) (India) · Mazovia (provider IP location) (Poland) · Setúbal (provider IP location) (Portugal) · Cluj County (provider IP location) (Romania) · Constanța County (provider IP location) (Romania) · Moscow (provider IP location) (Russia) · Gauteng (provider IP location) (South Africa) · Canary Islands (provider IP location) (Spain) · Madrid (provider IP location) (Spain) · Aargau (provider IP location) (Switzerland) · Geneva (provider IP location) (Switzerland) · Kamphaeng Phet (provider IP location) (Thailand) · North Holland (provider IP location) (The Netherlands) · Dnipropetrovsk Oblast (provider IP location) (Ukraine) · Kyiv City (provider IP location) (Ukraine) · Lviv (provider IP location) (Ukraine) · England (provider IP location) (United Kingdom) · California (provider IP location) (United States) · Florida (provider IP location) (United States) · Georgia (provider IP location) (United States) · Iowa (provider IP location) (United States) · Massachusetts (provider IP location) (United States) · Michigan (provider IP location) (United States) · Missouri (provider IP location) (United States) · Pennsylvania (provider IP location) (United States) · Texas (provider IP location) (United States) · Washington (provider IP location) (United States) · Ho Chi Minh (provider IP location) (Vietnam)
Countries
Australia, Belgium, Bulgaria, Czechia, Finland, Germany, Guatemala, India, Poland, Portugal, Romania, Russia, South Africa, Spain, Switzerland, Thailand, The Netherlands, Ukraine, United Kingdom, United States, Vietnam
Platform
API, CLI, Jupyter, SSH, custom Docker images, persistent volumesSDK: Go, JavaScript/TypeScript (official); Console REST API (managed wallet, API key)Terraform: community (terraform-provider-akash repo in the akash-network GitHub org, v0.0.x, not referenced in current docs)Kubernetes: no (workloads run as Kubernetes pods on provider clusters; no managed Kubernetes for tenants)Clusters: Multi-node Ray clusters advertised on the home page; confidential compute (Intel TDX / AMD SEV-SNP, NVIDIA Confidential Computing GPU pass-through) opt-in via SDL on capable providers
Storage
ephemeral · 21.99 USD/month for 256 GiB in the usage calculator default estimate (about 0.086 USD/GiB-month) · Estimate from the Akash Console pricing API shown in the calculator; the actual price is set in each provider bid. Data is lost on container restart.volume · 21.99 USD/month for 256 GiB in the usage calculator default estimate (about 0.086 USD/GiB-month) · Persistent storage classes beta1 (default), beta2, beta3 (NVMe, recommended); provider-local, survives restarts and updates on the same provider, lost on close or provider change. Price from the calculator page.
Migration out
Workloads are standard container images described in SDL, so they redeploy elsewhere unchanged. Persistent volumes are provider-local with no snapshot or export feature documented; data must be copied out (for example with the shell file-copy commands) before closing. Closing a deployment returns unused escrow; card credits already bought are non-refundable per ToS.

Security & compliance

SOC 2
not published
ISO 27001
not published
HIPAA
not published
GDPR / DPA
not published
Other
Some providers are marked 'audited'; tenants can restrict bids to audited providers with signedBy in the SDL (32 of 60 online providers audited in the providers API). Provider attributes may self-declare an uptime guarantee.
Legal entity
Overclock Labs Inc.
Funding
Open-source protocol with a native token (AKT) used for staking, governance and gas; stats page on 2026-09-29: 90.17M of 298.31M AKT bonded, inflation 4.00%
Read on the provider's own pages · 27 pages read · facts marked "not published" were looked for and not foundPages read ↓
Assessment · documents read Sep 29, 2026

Akash Network, in short

What stood out when we read the documents: strengths, weaknesses, who it suits, and what we could not find.

Strengths

  • Per-GPU-hour prices for 26 GPU SKUs (25 models) visible without an account, backed by a public no-auth API with min/max/median bid data
  • Low network prices: H100 SXM 2.72, A100 SXM 80GB 1.71, RTX 4090 0.30, RTX 5090 0.67 USD/GPU-h (bid-weighted averages)
  • Self-custody path needs no account and no KYC (docs: 'No KYC, permissionless'), funded in ACT/AKT
  • Card billing via Stripe with automatic escrow funding and optional Auto Top-Up in the managed Console
  • Container-native (SDL similar to Docker Compose), official Go and JS/TS SDKs, CLI and REST API
  • Opt-in confidential compute (TEE), including GPUs in NVIDIA Confidential Computing mode, on capable providers

Weaknesses

  • Thin supply: 429 GPUs of capacity network-wide per the stats page (446 total, 146 available in the gpu-prices API); several models have a single provider and zero available units
  • Persistent storage is provider-local and lost when a deployment closes or moves provider
  • No SLA, no status page, no published compliance certifications; ToS provides the services 'as is'
  • The price is only known after bids arrive; published figures are averages, not guaranteed rates
  • Crypto is no longer accepted in the managed Console; crypto users must self-host Console Air or use the CLI with a Keplr wallet
  • Card top-ups are converted to a stablecoin and are non-refundable per ToS
  • Egress not published; IPv4 is an extra IP lease

Ideal for

  • Cost-sensitive inference and batch jobs that tolerate provider variability
  • Crypto-native teams wanting permissionless, wallet-funded compute
  • Developers already packaging workloads as containers

Not published

  • Company HQ city/country and founding year not read on an official page on the check date (ToS names Overclock Labs Inc., governed by Texas law).
  • Egress/bandwidth pricing not published
  • set implicitly by provider bids.
  • Maximum GPUs per deployment and interconnect (NVLink/InfiniBand) not published at network level.
  • B200 has no data in the gpu-prices API and B300 has no price there
  • only the fixed rates on the pricing page are available, access terms not published.
  • Current protocol take fee on payments not read on an official page on the check date.
  • Actual trial credit amount: docs say 1 USD, marketing CTA says 100 USD.
  • Conditions that trigger identity verification: ToS section 2.2 reserves the right to verify identity and screen against sanctions lists
  • no threshold published.
  • Whether USDC is still accepted for self-custody escrow: deployments docs say ACT only (AKT under circuit breaker), the API docs page says AKT or USDC.
  • Regions are provider locations from IP geolocation in the Console providers API (online providers on 2026-09-29), not fixed data centres
  • cities not given.

Observations

GPU pricing page is public: 26 rows, sortable by VRAM, filterable by model, VRAM band and interface (SXM/PCIe); each row shows a 'Starting at' USD/hr figure and a 'Rent' button (B200/B300: 'Get access')

Footnote: 'Prices are per GPU-hour in USD, derived from recent provider bids on the network and weighted by each provider's GPU count. B200 and B300 are listed at fixed rates.'

Despite the 'Starting at' label, the figures equal the weightedAverage field of the public API console-api.akash.network/v1/gpu-prices, not the minimum bid (e.g. H100 minimum 2.04 vs 2.72 shown)

Home page comparison widget: H200 for 1 h, Akash $4.45/hr vs CoreWeave $6.31, AWS $7.91, Google Cloud $10.44; the same page claims 'H100s for $1.33/hr', inconsistent with the 2.72 on the pricing page

Usage Pricing Calculator prices CPU, memory, ephemeral and persistent storage separately (default 20 vCPU, 128 GiB RAM, 256 + 256 GiB: 305.78 USD/month) and compares with AWS Fargate, GKE and Azure Container Instances on a 730 h/month basis

Trial credit amount is inconsistent across official pages: docs 1 USD, blog CTA 100 USD, 2024 announcement 10 USD

Home page network stats: 60 active providers, 401 GPUs; stats.akash.network the same day: 62 providers, 429 GPUs capacity, 169 GPUs leased

Docs 'GPU Deployments' page still lists older price ranges (e.g. H100 2.50-4.00/hr, RTX 4090 0.50-1.50/hr) that differ from the live pricing page

Docs claim 'Providers in 85+ countries'; the providers API shows 60 online providers in 21 countries

Pages read

Questions about Akash Network

What GPUs does Akash Network offer?

Akash Network has 28 GPU types with a fresh offer today, for example H100 from $2.04, A100 80GB from $1.07, RTX PRO 6000 from $1.44, RTX 5090 from $0.57, RTX 4090 from $0.36, T4 from $0.11 per GPU-hour on-demand. The full list with prices and market medians is on this page.

How does Akash Network bill?

Billing granularity: per block (about 6 seconds, 600 blocks per hour) from an on-chain escrow. Minimum charge: No minimum rental period published. Self-custody deployments need an escrow deposit of about 5 ACT minimum plus AKT gas (about 0.02 to 0.1 AKT per transaction); Console Air ACT mint minimum 10 ACT. Stopped instance: No stop state: a deployment runs until closed or until escrow is depleted. Closing returns unused escrow; persistent storage is provider-local and is lost when the deployment is closed or moved to another provider.

How trustworthy is Akash Network?

Its Trust Score is 53/100 (v0.1), read from its own documents. Strong on company & track record. Weaker on infrastructure control, reliability signals, security & compliance. The score measures transparency and published guarantees, not customer satisfaction; every criterion and its source is listed above.

Where are Akash Network's data centres?

36 named regions in 21 countries: Australia, Belgium, Bulgaria, Czechia, Finland, Germany, Guatemala, India, Poland, Portugal, Romania, Russia, South Africa, Spain, Switzerland, Thailand, The Netherlands, Ukraine, United Kingdom, United States, Vietnam.

How do you pay Akash Network, and are there credits?

Payment by card, crypto. Free credits: Docs: 1 USD trial credit, no card, 30 days, deployments limited to 24 h each, GPUs up to RTX 3090 only. Marketing pages show other figures (blog CTA 'Start with $100 Free Credits'; Dec 2024 announcement 10 USD).. Identity check: ID on request.