Provider at a glance
- Category
- Decentralised Compute
- Entity
- Not verified
- Official website
- nosana.com
Independent rankingchecked How it works
Details
Scored 75 out of 100 under methodology 2026.10.18, from the evidence read in September 2026.
Nosana
GPU marketplace on Solana
Decentralised Compute
#2 Overall
Ranked #2 of 7 decentralised compute in the published ranking
75
ChainChoice Score
The ChainChoice Score, shown against 100: published scores run from 41 to 99.- High evidence confidence
Is Nosana right for you?
Right for you if these matter most
- Fetchable Price Per GPU-Hour8/10, the category best
- Verified vs Self-Reported Supply6/10; the category best is 7
Look elsewhere if these come first
- Published Utilisation and Live Availability7/10 here; Akash Network scores 9
- Recourse When a Provider Vanishes Mid-Job2/10 here; Akash Network scores 3
Not yet tested for your country. Markets it lists: Global.
to test where you can use it.
The rank above is the published one, the same for every reader. Find my best compute network shows where it lands for you.
What will Nosana cost you?
Not priced on this basis
- Usage per month
- 10
How safe is Nosana?
- Evidence: verified4 of 4 criteria rest on quoted, sourced receipts. Read September 2026.
- LicenceMiCA (EU): no register record on file here.
Alternatives to Nosana
Compare all 4- 1Akash Network4 points above · stronger on dc utilisation disclosure (9 against 7)79
- 3io.net13 points below · weaker on dc price transparency (5 against 8)62
- 4Fluence16 points below · stronger on dc failure recourse (3 against 2)59
For your setup
Find out if Nosana fits you
The rank above is the published one — the same for every visitor. Answer the category's questions and this section shows where Nosana lands for you, which criteria your answers weight up, and which stated needs the record does or does not meet. The published rank stays beside it.
- A few questions
- No account required
- Ranking-blind: no commercial input
- Personalised in under 60 seconds
Find my best compute networkNothing is inferred from clicks, popularity or commercial value.
Why this rank
Why Nosana ranks #2
Each strength is read from evidence on a 0–10 ladder and weighted as the published methodology states. The 32 objective points these produce (19.2 here) sit inside the score; availability, friction, trust record and measured data add the rest. Best in this ranking on 1 of 4.
4 published criteria · Σ weight 100
CriterionWeightStrength · 0–10ScoreContrib.MedianBestRankEvidence
308+7.685.08#1/7Loading evidence…
306+5.763.07#2/7Loading evidence…
207+4.483.09#2/7Loading evidence…
202+1.283.04#6/7Loading evidence…
Hairline = best in this ranking · tick = its median · a missing bar is "not scored", never zero
Margin−4behind Akash Network, #1
RobustnessHoldsacross a one-point change to each of its 4 criterion scores, the re-run scores 74–77 and stays #2
Evidence4 of 4 scored4 quoted, 4 with a source link
What would move the rank
Each line is the ranking run again with one of Nosana’s criterion scores changed and every other value held: the smallest whole-point change on that criterion that moves its rank, and where the re-run puts it. Up to 4 climbs and 3 falls are listed, smallest change first. Nothing here is a forecast.
Verified vs Self-Reported Supply 6 → 9 on its ladder · passes Akash Network+380#2 → #1
Recourse When a Provider Vanishes Mid-Job 2 → 6 on its ladder · passes Akash Network+480#2 → #1
Fetchable Price Per GPU-Hour 8 → 1 on its ladder · io.net passes it−762#2 → #3
Smallest climb: verified vs Self-Reported Supply 6 → 9 passes Akash Network, #1.
Strengths
2 from the score dataWhere it leads
Best for readers who weight
- Fetchable Price Per GPU-Hour — 8/10, the category best
- Verified vs Self-Reported Supply — 6/10, category best 7/10
Tradeoffs
2 from the score dataWhere it trails
Not for readers who put first
- Published Utilisation and Live Availability — 7/10 against a category best of 9/10
- Recourse When a Provider Vanishes Mid-Job — 2/10 against a category best of 4/10
Record watch-outs
- The flagship H100 tier is advertised at '$1.500/h' against '0 / 0 hosts' — the enterprise headline price has no supply behind it. At the 15 September 2026 read the A100 80GB and both Pro 6000 tiers had no free hosts either (A100 80GB '0 / 13 hosts', Pro 6000 (SOC2) '0 / 8 hosts', Pro 6000 Server Edition (CC) '0 / 16 hosts') and the 6000 Ada had one ('1 / 2 hosts'), so most free capacity sat on consumer cards such as the 3090 ('24 / 59 hosts') and 3080 ('11 / 29 hosts').
- Nosana's live table lists professional/datacentre cards WITH availability at fetch time: 'NVIDIA Pro 6000 (SOC2)' at '$1.000/h' with '5 / 8 hosts', 'NVIDIA Pro 6000 Server Edition (CC)' at '$1.100/h' with '4 / 4 hosts', and 'NVIDIA 6000 Ada' at '$0.700/h' with '1 / 2 hosts' — nine professional-tier hosts free. 'NVIDIA A100 80GB' is listed at '$0.900/h' with '0 / 13 hosts' (13 registered). RTX Pro 6000 Blackwell Server Edition is a 96GB datacentre part, not gaming hardware. Rewrite the caveat. Read again 15 September 2026: the same tiers read Pro 6000 (SOC2) '0 / 8 hosts', Pro 6000 Server Edition (CC) '0 / 16 hosts' and 6000 Ada '1 / 2 hosts' at unchanged prices, so the nine free professional-tier hosts no longer reproduce; the caveat above now says so.
- Incomplete selection that caused the wrong caveat. The page lists 21 models; the research omits 12, including every professional tier: 4000/A4000 $0.128/h (0/1), 5070 $0.150/h (11/32), 4080 $0.160/h (1/3), 5000/A5000 $0.200/h (0/0), 5080 $0.200/h (11/26), A40 $0.400/h (0/1), A6000 $0.400/h (0/1), A100 40GB $0.610/h (0/0), 6000 Ada $0.700/h (1/2), A100 80GB $0.900/h (0/13), Pro 6000 SOC2 $1.000/h (5/8), Pro 6000 Server Edition CC $1.100/h (4/4). All nine prices the research DID quote verified exactly, including H100 $1.500/h at '0 / 0 hosts'. Read again 15 September 2026: the page still lists 21 models and the pricing note now quotes 13 of them; host counts moved (5080 '0 / 29 hosts', Pro 6000 Server Edition (CC) '0 / 16 hosts'), and the 13 prices the note quotes match the August figures.
Competitive position
Where Nosana leads and where it trails
Rank on every objective dimension among the rows scored on it. A criterion in the top third of its pool is listed as a lead, the bottom third as a trail; the middle is neither. Popularity, size and commercial terms are not dimensions.
Availability#1 / 714.0 of 20 · constant here
Published criteria#2 / 714.4 of 30 · best 15.8
Friction#1 / 72.7 of 10 · constant here
Trust record#1 / 73.2 of 8 · constant here
Measured data#1 / 78.3 of 18 · constant here
Where Nosana leads
- Fetchable Price Per GPU-Hour#1 / 7
- Verified vs Self-Reported Supply#2 / 7
- Published Utilisation and Live Availability#2 / 7
Where it trails
- Recourse When a Provider Vanishes Mid-Job#6 / 7
Closest alternatives
The 3 nearest rows by score
Nearest by published score in either direction. Each card states the criteria on which the two rows differ and the cost difference under the same scenario inputs; the compare arithmetic below shows why the higher one ranks higher.
Akash Network#1 of 7 · score 79 · 4.0 points ahead
Nosana leads on
- Fetchable Price Per GPU-Hour8 vs 7
Akash Network leads on
- Verified vs Self-Reported Supply7 vs 6
- Published Utilisation and Live Availability9 vs 7
- Recourse When a Provider Vanishes Mid-Job3 vs 2
Largest single reason Akash Network is ahead: published criteria (+3.6 display points).Scenario cost: Akash Network has no verified pricing for this scenario.
io.net#3 of 7 · score 62 · 13.0 points behind
Nosana leads on
- Fetchable Price Per GPU-Hour8 vs 5
- Verified vs Self-Reported Supply6 vs 4
- Published Utilisation and Live Availability7 vs 4
io.net leads onNo scored criterion
Largest single reason Nosana is ahead: published criteria (+13.0 display points).Scenario cost: io.net has no verified pricing for this scenario.
Fluence#4 of 7 · score 59 · 16.0 points behind
Nosana leads on
- Fetchable Price Per GPU-Hour8 vs 5
- Verified vs Self-Reported Supply6 vs 3
- Published Utilisation and Live Availability7 vs 2
Fluence leads on
- Recourse When a Provider Vanishes Mid-Job3 vs 2
Largest single reason Nosana is ahead: published criteria (+16.1 display points).Scenario cost: Fluence has no verified pricing for this scenario.
Against the leader
79 vs 75 · gap 4.0Why Akash Network ranks above Nosana
The compare workspace's arithmetic applied to these two rows. Each dimension's share of the gap is its weight × the difference, on the displayed scale — it reproduces the displayed gap of 4.0 to within rounding.
DimensionThisAkash NetworkGap share
Availability · 23% weight · constant here14.014.0same
Published criteria · 35% weight14.415.8−3.6, worse
Friction · 12% weight · constant here2.72.7same
Trust record · 9% weight · constant here3.23.2same
Measured data · 21% weight · constant here8.38.3same
Published criterionThisAkash NetworkPoints
Fetchable Price Per GPU-Hour · w3087+0.96, better
Verified vs Self-Reported Supply · w3067−0.96, worse
Published Utilisation and Live Availability · w2079−1.28, worse
Recourse When a Provider Vanishes Mid-Job · w2023−0.64, worse
Net7579−4.0
Your scenario
Nosana is not priced for this scenario
Metered usage beyond whatever the plan includes, for a year. Excludes egress, support tiers and enterprise minimums. Change an input and every figure recomputes from the published rate; the published rank does not move.
your cost/yr · default scenario — set your own
Usage per month · your input10
Rate read from the provider’s published pricing ↗ on 15 September 2026.
your cost/yrCost is stated as your cost/yr in USD; it is never a yield, never annualised, and it does not enter the score.
Not priced on this basis
Under the default inputs · USDCategory medianNot priced0 of 7 rows priced
Cheapest pricedNone pricedDifference not stated · availability not applied — set your country to filter
7 of 7 rows in this ranking have no verified rate for this scenario and are absent from the median, not counted as free.
Evidence
Evidence read 15 September 2026
One row per published criterion: the page each strength was read from, its source type by domain, the reading date and whether the file and the published table agree. Open a row for the quote and the receipt.
4 of 4 cells scored
Score evidenceVerified4 of 4 criteria reconstruct from receipts
4 of 4 criteria reconstruct from a quoted, sourced receipt equal to the published strength.Public receiptsReading…4 quoted with a source URL
The receipt file has not been read yet.Evidence confidenceHighcoverage and corroboration, never freshness
4/4 criteria scored, with 4 dated sources quoted.RobustnessHoldsrank #2 across every one-point re-run
Whether the position holds when one criterion score is changed by one point and the ranking is run again. It is always a re-run of the ranking, never an estimate.The newest evidence read on the platform is 3 November 2026; this category's reading is older. The scores stand as read on 15 September 2026.
Reading the evidence ledger…
CriterionSourceTypeReadFreshnessStatusReceipt
Fetchable Price Per GPU-Hourstrength 8 · weight 30Report an error—15 September 2026*
Unknown
No reading date is recorded for this cell or its category.No receipt
A strength is published but no quote or source page is on file for it.Verified vs Self-Reported Supplystrength 6 · weight 30Report an error—15 September 2026*
Unknown
No reading date is recorded for this cell or its category.No receipt
A strength is published but no quote or source page is on file for it.Published Utilisation and Live Availabilitystrength 7 · weight 20Report an error—15 September 2026*
Unknown
No reading date is recorded for this cell or its category.No receipt
A strength is published but no quote or source page is on file for it.Recourse When a Provider Vanishes Mid-Jobstrength 2 · weight 20Report an error—15 September 2026*
Unknown
No reading date is recorded for this cell or its category.No receipt
A strength is published but no quote or source page is on file for it.How you can lose money here, and what the ranking scores
- Execution failure (Nosana 2/10): you get a worse price, later, or nothing.Scored by Recourse When a Provider Vanishes Mid-Job · 20%
- Information failure (Nosana 6/10): the number is wrong or bought, or your data leaks.Scored by Verified vs Self-Reported Supply · 30%
Not assessed by this ranking: Liveness failure, Exit failure. A high place says nothing about these; the Methodology tab says what each means. Also scored, as cost, quality or a summary of several families: Fetchable Price Per GPU-Hour (30%), Published Utilisation and Live Availability (20%).
Facts
facts, not scoresWhat the record publishes
- Pricing
- Public pricing table (fetched 15 Sep 2026): NVIDIA H100 '$1.500/h' — '0 / 0 hosts'; 5090 '$0.400/h' — '1 / 50 hosts'; 4090 '$0.320/h' — '0 / 34 hosts'; 3090 '$0.192/h' — '24 / 59 hosts'; 3080 '$0.096/h' — '11 / 29 hosts'; 4070 '$0.096/h' — '2 / 8 hosts'; 3070 '$0.080/h' — '7 / 14 hosts'; 4060 '$0.064/h' — '2 / 7 hosts'; 3060 '$0.048/h' — '5 / 19 hosts'; A100 80GB '$0.900/h' — '0 / 13 hosts'; Pro 6000 (SOC2) '$1.000/h' — '0 / 8 hosts'; Pro 6000 Server Edition (CC) '$1.100/h' — '0 / 16 hosts'; 6000 Ada '$0.700/h' — '1 / 2 hosts'. 21 models listed. Docs: prices are per-second and set per GPU market, readable unauthenticated from api.nosana.com/api/markets/. Claims 'savings of up to 6× depending on workload'. Explorer: 'Online GPU Hosts 1,018', 'Running jobs 921' (live figures); the '2,000 nodes' line on the homepage is a customer testimonial.Nosana — official pricing page ↗ · 15 September 2026
- Prices published in
- USDNosana — official pricing page ↗ · 15 September 2026
- Licences held
- Not published
- Compliance level
- partial
- Regulatory note
- Not published
- Supported chains
- Multi-chain
- Markets served
- global
- Primary markets
- Not published
- Restricted markets
- Not published
- Operating limits
- Not published
- Availability note
- Terms (last updated 3 Aug 2026) bar anyone subject to sanctions or located in a sanctioned country or territory, without naming countries. Governed by the laws of the Netherlands.
No confirmed legal entity is on file for this record. Nothing is inferred from the brand name.
If something goes wrong with Nosana: where to complain, and where to report a scam
Loading the routes for your country.
How this page is reviewed
Reviewed means this product in this category was read against its sources on or after that date; every ranked category is re-read at least every 45 days. Models research, extract and parse. A deterministic engine ranks. Every quote passes an adversarial refute pass.
Accountable for this review: Mathias Siemonsmeier, Founder.
How we review · Review calendar · Corrections log · How we make money
History
What the record carries
Changes to published ranks are recorded in a ledger; every step and its cause is listed on the rank-history page. The entries below begin with the current verified snapshot, then the methodology versions whose changes applied to this category, newest first.
8 methodology entries
- 15 September 2026Current verified snapshot · methodology 2026.10.18Score 75, #2 of 7. 4 of 4 criteria scored. History starts here; earlier values are not interpolated.
- 10 October 2026
2026.10.8When nothing serves your country, the list shown for reference is the published ranking · platform-wide- fixed The reference list shown when nothing serves your country is the published order with the published scores, and a build check now fails any such list out of that order or with a score the ranking does not publish.
- fixed The explorer prints an overall rank only for a row the published ranking holds; a country entrant such as Coinbase Commerce for US merchants shows no overall rank.
- changed Published orders do not move. Country results change only where nothing serves the country: in India bill pay keeps Bitcoin Well first and now shows Strike second, as published; 35 such runs across all categories and countries now match the published list exactly.
- 10 October 2026
2026.10.6A criterion you require re-scores the rows that meet it · platform-wide- changed The explorer sends each criterion you require, at any band, as fit_required with your other answers; region, licence, chain and fact filters stay filters. On /best, a required criterion joins For you as a marked preference does, which needs your country; Overall stays the published ranking, filtered.
- added Every published criterion has a golden case as the explorer sends it (require:<criterion>), so the order a requirement gives is replayed and gated as the published order is, and a change to it needs an entry here.
- 16 September 2026
2026.09.15Dimension scores carry tenths, and eleven more categories are re-read from primary sources · platform-wide- changed Dimension scores resolve to tenths of a point (O-04b), and weightedTotal is rounded to six decimal places before the order is read. 518 published scores move across 107 categories that were not re-read; none of them moves by more than 4 points, and most move by 1.
- fixed 66 published scores change in the eleven re-read categories, against a verbatim source each. The largest moves are Reflexer RAI (−16), Beefy and Unrekt (−11), Aztec (−10) and Golem (+9) — all of them rows whose sources had moved since their last read.
- changed Published order moves in 38 of 119 categories and three #1s change: smart_contract_audits (Zellic → Certora) and travel_rule_compliance (Sumsub Travel Rule → CodeVASP) on the finer resolution, kyc_compliance (Veriff → Sumsub) on the re-read.
- added availabilitySource (O-04g): for a reader who states a country, a row whose availability in that country rests on a region tag its record gives no source for sorts below every sourced row. It is inert in every order published here, which states no country.
Methodology 2026.10.18, the same published run every visitor sees · receipt decentralised_compute:nosana_dcompute:2026.10.18:9c837a03b7fd
Are you Nosana? Reply to this verdict, or send evidence, ask for a correction or claim the listing. Every message gets a public ticket number; none of it changes a rank unless a public page or a signed statement supports it.