AI Patent Search Tools in 2026: The Market, the Features and the Prices
2026-11-20 · Mohammed Mujahid
We counted 58 AI-assisted patent search tools, and only 14 publish a full price list. The market by segment, what the tools cost, what changed in 2026 and how to choose.
We counted 58 AI-assisted patent search tools on the market in November 2026, and only 14 of them publish a full price list. Half (29) will only quote you a price after a sales call.
The category has changed fast. Two years ago, "AI patent search" mostly meant a semantic-search box bolted onto a Boolean database. Today the same label covers agent swarms that read 180 million patents for a litigation team, free chat-style tools anyone can use, and full platforms that move from prior-art search to claim charts to a first draft.
For patent professionals, that creates a real buying problem. The tools are hard to compare, the prices are mostly hidden, and vendor benchmarks are not independent. This article lays out what we found: who the players are, what they actually do, what they cost, how we think buyers should choose, and why expert judgment still decides the result.
Key findings
- 58 tools in 7 segments, from enterprise databases (Derwent, Orbit, Patsnap, PatSeer) to litigation agents (&AI, Stilta) and free public tools (Perplexity Patents, WIPO PATENTSCOPE).
- Pricing is mostly opaque. 29 tools are quote-only; 14 publish full prices; 9 publish only part of their pricing.
- Entry prices span three orders of magnitude, from free and $20 a month to third-party estimates of $24,000 to $72,000 a year for a team license.
- Credits are replacing seats in newer tools, so cost tracks the number of searches and charts you run.
- 2026 was a year of agents: Questel, Clarivate, LexisNexis and WIPO all shipped new AI assistants or agent platforms, and litigation-focused startups raised large rounds.
How we built the scan
We included any tool that uses AI (semantic retrieval, language models or agents) to find or rank patent prior art. We also kept the main free public databases as a baseline, even where their AI is thin.
Everything here comes from public sources only. We started from vendor pricing pages, launch announcements and product documentation, then cross-checked against independent comparison guides and software directories. We then compared our list line by line with a second, independent AI-assisted research pass. We added the tools it found and kept only claims we could trace to a public page.
This snapshot is dated 20 November 2026. Prices, features and vendor benchmarks change often; the disclaimer at the end sets out the limits.
The market in seven segments
Low-cost self-serve tools and AI-native search specialists tie as the largest groups, with the enterprise databases close behind.
Self-serve / low-cost, 12
AI-native search, 12
Enterprise search & analytics, 11
Lifecycle AI (search + drafting), 9
Free / public, 7
Litigation / claim charts, 6
R&D intelligence, 1
Segment · What it is · Examples
Self-serve / low-cost (12) · Sign up online, pay monthly or per report, often with a free tier · PQAI, FindIP, GoVeda, Patentcloud
AI-native search (12) · Built around semantic, graph or agentic AI retrieval from a description or a patent · IPRally, Amplified, Clerq, Patentfield
Enterprise search & analytics (11) · Large curated databases with Boolean depth, legal status and analytics, now with AI layers · Derwent, Orbit, Patsnap, PatSeer
Lifecycle AI: search + drafting (9) · One platform from prior-art search to claim charts and drafting · Patlytics, Solve Intelligence, DeepIP, XLSCOUT
Free / public (7) · Patent-office databases and free AI tools · Perplexity Patents, WIPO PATENTSCOPE, Google Patents, Espacenet
Litigation / claim charts (6) · Agentic invalidity and infringement search with litigation-grade charts · &AI, Stilta, Garden, ClaimHit
R&D intelligence (1) · Patents mixed with papers, grants and market data for R&D teams · Cypris
The full list of all 58 tools, with links and pricing, is in the comparison at the end of this article.
The lines between segments are blurring. Enterprise vendors are adding agents and drafting, and lifecycle startups are moving into litigation. Expect consolidation: Clarivate already owns incoPat and Rowan Patents, and DeepIP bought PatentMaker this year.
What pricing really looks like
If a tool shows its price, it is almost always a self-serve tool under $500 a month. Enterprise platforms almost never publish a figure. Of the 58 tools, 29 are quote-only, 14 publish full pricing, 9 publish part of it (a free tier or a starting price), and 6 are free public services.
Published entry prices run from free (PQAI, FindIP, Patentfield, GoVeda and Patsnap Eureka all have free tiers) up to about $500 a month (Amplified, billed annually). For quote-only platforms, the only public numbers are third-party estimates, which run from several thousand to tens of thousands of dollars a year for a team. Every tool's pricing and trial terms are in the full comparison at the end.
The bigger shift is credits. Patsnap, GoVeda and FindIP charge by search credit. &AI moved to self-serve credit pricing in July 2026, arguing that patent work comes in bursts, so cost should follow the searches and charts actually run. One Transform took the opposite route: one flat monthly fee for unlimited searches.
One practical tip: when comparing credit plans, count what each task costs. On Patsnap's Pro plan, 10 credits buy 10 novelty searches but only 5 FTO searches.
Watch for a third pricing model. Seats and credits both charge for access to a tool. Some buyers, especially smaller teams, would rather pay for the outcome: an expert who already runs these tools and returns checked, usable work. We expect more offers built that way in 2027, ours included.
What changed in 2026
Three threads ran through the year. Incumbents put AI agents and assistants on top of their curated data. Litigation-focused startups raised large rounds. And prices moved toward credits and self-serve plans.
Date · Tool or company · What happened
7 Nov 2026 · USPTO Patent Public Search · Sign-in with a USPTO account now required; the public tool still has no semantic search
Sep 2026 · Wert Intelligence (Keywert) · Showed AI chemical-structure search at PIFC Tokyo; selected by LG Electronics
18 Aug 2026 · Clerq (formerly NLPatent) · Rebranded and launched agentic patentability workflows that return cited reports in about 10 minutes
18 Aug 2026 · AuriQ Patent Analysis AI · Beta launch with a free tier: prior-art search, claim charting, drafting
29 Jul 2026 · One Transform · Relaunched with one flat monthly fee for unlimited searches
Jul 2026 · WIPO PATENTSCOPE · Added a free AI assistant that turns plain language into structured queries
Jul 2026 · &AI · Moved to fully self-serve, published credit pricing
Jul 2026 · IPRally · API and MCP connector in preview
Jun 2026 · The Lens · Sold by Cambia to a non-profit owned by SPIE; free accounts now limited to academics
Jun 2026 · DeepIP · Acquired PatentMaker, adding a Munich office
29 May 2026 · Clarivate IPOne · AI agents and MCP connectors across Derwent, Darts-ip and CompuMark data
19 May 2026 · Stilta · $10.5M seed led by Andreessen Horowitz for agentic invalidity and infringement analysis
13 May 2026 · LexisNexis Protégé · AI assistant in PatentSight+ generally available
27 Apr 2026 · Questel QaECTER · New patent-search model deployed inside Sophia Search and Orbit
Apr 2026 · Patlytics · $40M Series B led by SignalFire
15 Apr 2026 · ClaimHit · Launched 60-second infringement screening from India
Mar 2026 · PatSeer · Integrated non-patent literature search; easier Design AI image search
Mar 2026 · DeepIP · $25M Series B
The launch that set the backdrop came late in 2025: Perplexity Patents, a free natural-language patent research agent, opened worldwide on 30 October 2025. Free, good-enough first-pass search is now the baseline that every paid tool has to beat.
How to choose, and where experts still matter
Start from the job, not the tool. A patentability screen, an FTO clearance and an IPR invalidity search need different coverage, different recall and different evidence standards. Most teams will end up with two or three tools, not one.
A short buyer checklist:
- Match the segment to the task. Litigation tools (&AI, Stilta, Garden) are built for invalidity and evidence-of-use charts. Enterprise databases (Derwent, Orbit, PatSeer, Patsnap) carry curated family, legal-status and classification data that FTO work depends on.
- Check coverage where your art lives. For Chinese, Japanese or Korean art, look at incoPat, Patentics, Patentfield, Keywert or PATENTSCOPE's cross-lingual search, not only English-first tools.
- Ask for non-patent literature. Products, standards, papers and archived web pages often decide invalidity. Several newer tools search them; many databases still do not.
- Run your own benchmark. Take three closed matters where you know the best references and see which tool finds them, and how high it ranks them. Vendor recall figures come from vendor test sets.
- Demand traceability. Every reference and claim-chart cell should link back to a dated, citable source an examiner or a court will accept.
- Price the work, not the seat. Estimate your monthly searches by type and price them in credits; a cheap seat can be expensive per FTO search.
- Check confidentiality terms. Invention disclosures are sensitive. Confirm how inputs are stored, whether they train models, and where they are hosted.
- Decide who checks the output. Every tool above needs a trained reviewer. If you do not have one in-house for every technology or every type of search, plan for expert support before you plan for another license.
AI tools now find good references faster than any human can read them. What they cannot do is decide which reference matters, how two documents combine for obviousness, or whether a claim chart will survive cross-examination. That judgment still needs an expert.
We think that is where the market goes next. Buying one more license will matter less than having an AI-powered domain expert on call: someone who runs these tools every day, checks every result, and hands back work your team can use straight away. At M-Son IP we are preparing a simple way to have our experts work alongside your team in exactly that way, across patent research. We will share more soon.
Full comparison: all 58 tools
Every tool in the scan, grouped by segment. "Quote" means the vendor publishes no price. Figures marked as third-party are estimates from review or directory sites, not vendor prices. Prices are in USD unless stated. Most names link to the vendor's own site; where we could not confirm the official site from public sources, the link goes to the public page we relied on instead.
Tool · Segment · Pricing · Free plan or trial
FindIP · Self-serve · Free (30 credits/mo); Basic $45/mo; Pro $135/mo · Free plan
GoVeda · Self-serve · Free; Pro from $200/mo (20,000 credits) · Free plan; 14-day Pro trial
One Transform · Self-serve · One flat monthly fee, unlimited searches (amount not published) · 7-day trial
Patent Inspiration · Self-serve · From €200/user/mo · Free trial listed
Patent77 · Self-serve · Free; from $49/mo · Free plan
Patenta.ai · Self-serve · Solo $299/mo · Free trial listed
Patentcloud · Self-serve · Free; from $19/mo (third-party listing) · Free version and trial
Patently · Self-serve · Free; $99 to $249/mo (directory listings) · Free plan
PatentScan.ai · Self-serve · About $99/report (third-party; confirm with vendor) · First report free
PQAI · Self-serve · Free; PQAI+ $20/mo; API from $20/mo · Free plan
PriorArtCheck · Self-serve · $149/report · None
SenseIP · Self-serve · Free version; paid tiers not published · Free version
AI Samurai · AI-native search · Quote · On request
Ambercite · AI-native search · Quote · 21-day trial (restricted)
Amplified · AI-native search · From $500/mo, billed annually · Free trial listed
Clerq (formerly NLPatent) · AI-native search · Quote · Free trial
DorothyAI · AI-native search · Quote · Demo
IPRally · AI-native search · Quote (third-party estimate for teams: $24,000 to $72,000/yr) · 3-day trial
Keywert · AI-native search · Annual subscription; quote · None listed
Octimine · AI-native search · Quote (third-party estimate: about $8,000 to $10,000/yr for 1 to 3 users) · None listed
Patentfield · AI-native search · Free; Basic $100/mo · Free plan
Patentics · AI-native search · Quote · None listed
Quartet · AI-native search · Quote · Demo
Relecura · AI-native search · Free plan; paid from $9 (third-party listing) · Free plan and trial
AcclaimIP (Anaqua) · Enterprise · Quote · None listed
Derwent Patent Search / IPOne · Enterprise · Quote · None listed
IamIP · Enterprise · Quote · None listed
incoPat · Enterprise · Quote · Free trial
InnovationQ+ (IP.com) · Enterprise · Basic $49/mo; Premium $199/mo; Day Pass $199 · Free trial listed
LexisNexis TotalPatent One / PatentSight+ · Enterprise · Quote · None listed
Minesoft Origin / PatBase · Enterprise · Quote · None listed
Orbit Intelligence (Questel) · Enterprise · Quote · Demo on request
PatSeer · Enterprise · Quote (third-party estimate: about €100 to €500/mo) · 2-week free trial
Patsnap Eureka · Enterprise · Free Basic; Pro Searching $400/mo; Enterprise quote · Free tier; 14-day trial
WIPS Global · Enterprise · Quote · None listed
Ankar · Lifecycle AI · Quote · Demo
AuriQ Patent Analysis AI · Lifecycle AI · Free tier in beta; paid tiers not published · Free tier
DeepIP · Lifecycle AI · Quote · Trial on request
IP Author (Dolcera) · Lifecycle AI · From $499/mo (third-party listing) · None listed
IP Copilot · Lifecycle AI · Quote · Demo
Patdel Analytics · Lifecycle AI · Not published · None listed
Patlytics · Lifecycle AI · Quote (third-party estimate: $800 to $2,000/user/mo) · Demo
Solve Intelligence · Lifecycle AI · Quote (third-party estimate: about $9,300/user/yr) · Demo
XLSCOUT · Lifecycle AI · Quote · Trial listed by some directories
Espacenet (EPO) · Free / public · Free · Free
Google Patents · Free / public · Free · Free
PatentAssist search · Free / public · Free, no sign-up · Free
Perplexity Patents · Free / public · Free in beta; higher limits on paid Perplexity plans · Free
The Lens · Free / public · Free for non-commercial use; commercial license $1,000/yr · Free (non-commercial)
USPTO Patent Public Search · Free / public · Free (USPTO account sign-in required) · Free
WIPO PATENTSCOPE · Free / public · Free · Free
&AI · Litigation · Pay-as-you-go credits; Pro about $625/mo (third-party tracker); Enterprise quote · 7-day team trial
ClaimHit · Litigation · Starter $199 one-time; Growth $449/user/mo · 3 free searches
Garden · Litigation · Quote · Demo or waitlist
PatentWatch · Litigation · From $50/mo (third-party listing) · None listed
Stilta · Litigation · Quote · Demo
Traindex · Litigation · Free Patent Bot; enterprise quote · Free Patent Bot
Cypris · R&D intelligence · Quote · None listed
Want help choosing or benchmarking tools for your own matters, or would you rather have an expert run them for you? We are opening early conversations about a new way to work with M-Son IP's AI-powered experts. Ask an expert. Free scoping, a reply within one business day, and an NDA before details.
Sources
Main sources (accessed November 2026):
Disclaimer
This article is for general information only. It is based solely on publicly available information (vendor websites, press releases, product documentation, software directories and third-party review sites) gathered up to 20 November 2026. We did not contact vendors, receive vendor input, or test the tools hands-on. Sources are linked throughout the article.
Features, prices and plans change often and may differ by region, contract or customer. Prices marked as estimates come from third-party sites, not from the vendors. Benchmark and performance figures quoted are the vendors' own claims and have not been independently verified. Please confirm all details with each vendor before buying.
Inclusion or omission of a tool is not an endorsement or a criticism, and the order of tools carries no ranking. Nothing here is legal advice. Product names and trademarks belong to their respective owners. This article is independent: no vendor sponsored it, paid to be included, or reviewed it before publication. If you spot an error or an outdated detail, please tell us and we will correct it.