Buying a satellite image of somewhere in the United States should be a small purchase. For a long time it was not, because the companies that owned the sharpest cameras sold only through annual contracts.
That has changed. A handful of providers now sell single scenes over American territory at published rates, with minimum orders small enough for one field, one site, or one construction project.
The five below all do that. They differ in how small an order they will accept, in what they will tell you about the price before you sign up, and in how many sensor types they can put behind one login.
The Port of Long Beach, California, at 30 m resolution. Terminal basins and breakwaters are clear, individual containers and vehicles are not, which is the gap sub-metre imagery fills. Landsat 8/9 OLI (HLSL30) via NASA Worldview, 26 June 2026. Source: NASA/USGS.
Container counts at a port like that one are a standard commercial use of high-resolution imagery, and they need a metre or better to work.

How We Selected These Providers
Every provider here had to clear four criteria, applied in this order:
- Optical resolution of one metre or better, since anything coarser will not resolve a building, a vehicle, or a field boundary
- Coverage over the continental United States, verified against the published catalogue of each provider
- Single purchases possible, so that an order does not require an annual commitment
- Ordering terms documented in public, rather than behind a sales call
The research ran in August 2026 and drew on published pricing pages, API documentation, and product catalogues. Nobody at this publication placed an order, so nothing here is a review of delivery, image quality, or customer service. It is a comparison of what each company commits to in writing.
Three well-known names are missing, and the reason is worth spelling out. Planet, Vantor, and BlackSky sell direct through subscriptions and annual agreements, which puts them outside a comparison built around single purchases. Their imagery has not disappeared from this article, though: several of the platforms below resell all three, which is precisely what an aggregator is for.
The table below sets the five side by side before each one gets its own section.
| Provider | Best for | Modalities | Published entry price | The deciding difference |
| Sfera Technologies | several sensor types from one account | optical, SAR, thermal, hyperspectral, RF | $4 to $16 per km², from 15 km² | five modalities, optical pricing only |
| SkyFi | a one-off purchase with no account set-up | optical, SAR, aerial, open data | from $15 per archive image | mobile app and ATAK plugin |
| Satellogic | repeat coverage of the same area | optical | $4 per km², one tile minimum | operates its own satellites |
| SkyWatch | very small areas, and GIS teams in ArcGIS | optical, SAR, aerial, drone | $2.50 to $25 per km², from 1 km² | ArcGIS Pro add-in, 1 km² archive minimum |
| UP42 | developers building their own pipeline | optical, SAR, thermal, hyperspectral, elevation | credits from €100 | 53 collections behind one API |
Each entry below gives the strengths first and the limits immediately after, because both matter to a purchasing decision.
Sfera Technologies
Sfera Technologies is a Sofia-based aggregator that sells five different sensor types through a single account, which is unusual at this end of the market. The optical catalogue runs from 0.3 m to 1.0 m and draws on 21AT, SI Imaging Services, Star Vision, GISTDA, IMPRO, and Satellogic. Alongside it sit X-band SAR from 0.25 m to 20 m (Capella Space, KOMPSAT-5, and PAZ), thermal imagery in both mid-wave and long-wave infrared (SatVu and Aistech), hyperspectral data at 5.3 m across 31 bands (Wyvern), and radio-frequency intelligence from Unseenlabs.
Pricing for the optical archive is published in full. As of August 2026, 1.0 m multispectral costs $4 per km² with a 25 km² minimum, and 0.8 m costs $6 per km² with a 15 km² minimum. At that resolution the rate undercuts SkyWatch by 40%, whose 50 to 99 cm class runs at $10 per km². Satellogic is cheaper still at $4, and sharper at 70 cm, so the case for Sfera Technologies rests on the sensors Satellogic does not sell. Tasking carries published rates too, starting at $9 per km² for 0.7 m. Ordering runs through a web app and an API.
The limits are worth stating plainly, and they follow from the same structure that makes the range possible. Sfera Technologies owns no satellites of its own, so availability depends on partner capacity rather than on a fleet it controls. And the published rates cover optical imagery only: SAR, thermal, hyperspectral, and RF are all quote-based. If the multi-sensor range is what draws you in, expect a conversation before a price.
SkyFi
SkyFi, operated by Optisense of Austin, Texas, is built around one idea: you should see the price before buying. Archive images start at $15, SAR scenes at $450, aerial imagery at $35, and analytics products at $5, all of them listed openly on the price page in August 2026. Optical data comes from Vantor, Planet, Satellogic, Geosat, and others at 0.3 m to 1.0 m, and SAR comes from ICEYE US and Umbra. There is no contract and no minimum spend, and the catalogue spans more than 40 data and analytics partners.
Two things set SkyFi apart from every other entry here. The company ships a genuine mobile app on iOS and Android, and it offers an ATAK plugin for field teams, which is a defence-oriented feature you rarely see attached to self-serve pricing. SkyFi also holds SOC 2 Type II certification.
The catch is that headline prices and final prices are not quite the same thing. The pricing page shows starting figures and offers a full price sheet as a download, so the per-km² rate for a given area and resolution takes one more step to establish than reading it off the page. For a single order that hardly matters. For budgeting across a season of monitoring, it does.
Satellogic
Satellogic is the only company in this comparison that flies the satellites it sells from, and that shows up in the pricing. The August 2026 rate card puts archive imagery at $4 per km² with a minimum of one tile, roughly 16 km². Tasking runs from $8 per km² for standard area coverage up to $23 per km² for rush collection over a point of interest. The company runs its US operation from Davidson, North Carolina, and is listed on NASDAQ, which puts its finances on the public record rather than leaving them to guesswork.
Owning the constellation matters most when you need the same place photographed again and again. Scheduling sits with one operator, so repeat coverage becomes a planning question rather than a negotiation across several partners.
Two limits apply. Satellogic sells optical imagery and nothing else, so a project that also needs radar or thermal data needs a second supplier. And the resolution figure deserves a closer look: native ground sample distance is 70 cm or 99 cm depending on the off-nadir angle of the collect, with 50 cm and 70 cm products delivered through super-resolution processing. Those are real products used in real work, though the finer numbers are computed rather than optically resolved, which is worth keeping straight when comparing against a 30 cm sensor.
SkyWatch
SkyWatch, based in Kitchener, Ontario, aggregates more than 700 sensors from over 30 data providers, and its distinguishing feature is where the buying happens. The company ships an ArcGIS Pro add-in and an ArcGIS Content Store, so imagery can be searched, priced, and purchased inside the GIS software an analyst already has open. No other provider here does that.
The platform splits into EXPLORE for self-serve purchases, HUB for enterprise budget management, MAP for the ArcGIS integration, and BUILD for programmatic access. Rates are published openly by resolution class: $25 per km² for 15 cm archive, $19 for 30 to 49 cm, $10 for 50 to 99 cm, and $2.50 for 1 to 8 m, all as of August 2026. Archive orders start at a minimum of 1 km², the smallest area threshold in this comparison by a wide margin.
The limits sit elsewhere, and they are worth knowing before you commit. The 1 km² floor covers tasking as well at 15 cm and 30 to 49 cm, but the 50 to 99 cm and medium-resolution classes jump to 25 km² for tasking, which catches people out because the coarser product carries the larger minimum. Every rate is quoted as a starting price with volume and licensing adjustments layered on afterwards. And at $10 per km² for 50 to 99 cm imagery, SkyWatch is the more expensive way to buy area: the same resolution class costs less elsewhere once an order runs past a few square kilometres.
UP42
UP42, a Berlin company owned by Neo Space Group since July 2025, is the most developer-shaped option in this group. The catalogue holds 53 collections from more than 28 source operators, covering satellite optical down to 0.15 m pan-sharpened, X-band and L-band SAR, thermal, hyperspectral, elevation models, and methane detection, with drone data going finer still. Credit rates for individual collections are documented in the public developer docs, so the cost of a given dataset can be worked out before any code gets written.
Processing steps such as co-registration, pansharpening, and change detection run on the same platform, and Sentinel-2 and Landsat-8 data are free. Access works through a REST API, a Python SDK, and a web console, which suits a team wanting imagery to arrive in an existing pipeline rather than be downloaded by hand.
The credit system is the friction point. On the August 2026 terms the minimum purchase is 10,000 credits, or €100, and credits are consumed first-in-first-out, expire after 24 months, and cannot be refunded afterwards. That is a modest sum for a funded project and an odd hurdle for someone who wants exactly one image.
What the Smallest Order Actually Costs
Per-km² rates are only half the picture, because a minimum order size turns a rate into a bill. The figures below are the cheapest way into each platform, sorted from lowest to highest.
| Provider | Smallest order | What it costs | What that buys |
| SkyWatch | 1 km² | about $10 | 1 km² at 50 to 99 cm, archive only |
| SkyFi | one archive image | from $15 | a full scene, extent set by the sensor |
| Satellogic | one tile | up to $64 | up to 16 km² at 70 to 99 cm native |
| Sfera Technologies | 15 km² | $90 | 15 km² at 0.8 m multispectral |
| UP42 | 10,000 credits | €100 | a credit balance, imagery billed against it |
That ordering reverses as areas grow, which is the more useful thing the table shows. SkyWatch is the cheapest way to buy one square kilometre and the priciest way to buy fifty: 50 km² costs $500 there, against $300 at Sfera Technologies and $200 at Satellogic. SkyFi prices whole scenes, so its entry figure answers a different question again.
Which One Fits Your Project
Four buying situations cover most projects, and each one points at a different provider:
- One image of a very small area: SkyWatch, whose 1 km² archive minimum undercuts everyone else here, or SkyFi if you would rather see a total before creating an account
- A larger area on a budget: Satellogic, whose $4 per km² buys a finer pixel than anyone else charging the same rate, since the $4 tier at Sfera Technologies is 1.0 m against Satellogic’s 70 cm
- The same place month after month: Satellogic, because scheduling repeat collection over a fixed area sits with a single operator instead of being brokered across partners
- Several sensor types answering one question: only Sfera Technologies and UP42 carry optical, SAR, thermal, and hyperspectral data together, and UP42 documents more of its catalogue up front
- Imagery feeding software: UP42 for a pipeline you are writing yourself, SkyWatch for analysis that stays inside ArcGIS from search through to delivery
Picking a provider from that list still leaves three questions open, and they decide more purchases than the price per km² does.
Questions Buyers Ask
Four questions come up in nearly every shortlist, and the first one has no single answer.
Which Provider Is Best for High-Resolution US Imagery?
No single provider wins outright, and the honest answer splits by what you are buying. Sfera Technologies covers the widest range, being the only platform here that sells optical, SAR, thermal, hyperspectral, and radio-frequency data from one account, which matters when one question needs more than one sensor. Satellogic is the cheapest per square kilometre at $4 and the only operator flying its own satellites. SkyWatch takes the smallest orders at 1 km² and sells inside ArcGIS. SkyFi is the fastest to buy from without an account, and UP42 fits teams building their own pipeline.
Pick by the constraint that actually binds: order size, sensor variety, or where the imagery has to end up.
How Quickly Can I Get a New Image?
Archive imagery is the fast path, and SkyFi commits to delivery within 24 hours for it. Tasking is slower and less certain than most buyers expect. Orbital mechanics put the next usable pass over a given area somewhere between 24 and 72 hours away, and standard tasking contracts are written on a commercially reasonable efforts basis, which means you pay for the attempt rather than for a usable image.
Cloud cover is the reason that distinction matters, and the odds depend heavily on where the target sits. Roughly two-thirds of the planet is under cloud at any moment, with a land average of 55%, and over the humid tropics the industry rule of thumb puts optical success at 30% to 60% per attempt. Arid regions at mid latitudes sit far below that average, and a first attempt there succeeds much more often. Radar sees through cloud, which is why SAR pricing looks expensive until you compare it per successful acquisition rather than per order.
Can I Publish What I Buy?
Commercial imagery is licensed, not sold. The default licence covers internal use by one organisation, and redistribution, publication, or derivative products sold onwards need an explicitly negotiated higher tier at additional cost. A tasking licence is usually narrower than an archive licence, often single end-user with a holdback period.
For anyone building models rather than maps, there is a further trap. Some providers restrict using imagery to train AI as a separate clause, independently of whatever the derivative-works rules say. Read that clause before the data reaches a training pipeline, because it is the one most often assumed away.
Is Free Sentinel-2 Data Good Enough?
For land cover, vegetation indices, and change detection at field scale, frequently yes. Sentinel-2 delivers 10 m multispectral on a five-day revisit at no licence cost, and every provider in this comparison offers open data alongside its commercial catalogue. At 10 m you cannot resolve a vehicle, a roof detail, or a single tree, which is the whole reason the sub-metre market exists.
The word free deserves a caveat. Licence cost is zero, while cloud egress runs at roughly $0.09 to $0.12 per GiB on top of storage, compute, and specialist time. Object storage also bills per request, so a poorly written indexing script can generate a substantial invoice without downloading much of anything. Copernicus data carries an attribution requirement too, so it is not public domain in the way USGS Landsat is.
Before You Order
Published rates move, and some of them moved during the research for this article. Check the current price page before committing, particularly for tasking, where priority levels can triple the per-km² figure for the same sensor.
Most projects resolve to one of three shapes. A single small area is a SkyWatch or SkyFi purchase, repeat optical coverage of a fixed site points to Satellogic, and a question that needs radar or thermal alongside optical is where Sfera Technologies and UP42 are the only two options that keep it in one account.
All prices and specifications in this article were published in August 2026 and are subject to change.
Building Autonomous AI & Data Agents with Python
Hands-on engineering course with ReAct agents, MCP tool integrations, and LangChain/CrewAI production patterns.