AI Skills Assessment Platforms Compared: True Per-Candidate Cost Across HackerRank, Codility, TestGorilla, Vervoe, iMocha, and Plum (2026)

Before you compare prices you have to compare categories, because 'AI skills assessment' is doing heavy lifting across three products that barely overlap. **HackerRank** and **Codility** are coding-interview platforms. Their core asset is a sandboxed code editor with a test runner, a library of programming problems, and a pair-programming room (CodePair, CodeLive). The AI layer is mostly cheat detection — plagiarism scoring, webcam proctoring, signal analysis — plus newer AI question generation. If you are not hiring software engineers, neither of these is your tool. HackerRank publishes plans starting at roughly $100/month for Starter and $250/month for Pro at https://www.hackerrank.com/products/main/plans/.

**TestGorilla** and **iMocha** sit in the catalog-of-tests category. The product is a library — 400+ tests at TestGorilla, 3,000+ at iMocha — covering cognitive ability, personality, role-specific knowledge (Excel, SQL, Java, customer service, accounting), and language proficiency. You assemble a multi-test assessment, send a link, and get a candidate scorecard. The AI layer is anti-cheat (browser lockdown, randomization, webcam) and increasingly AI-authored questions and AI skill simulators. TestGorilla starts at $33/month (https://www.testgorilla.com/pricing/); iMocha is roughly $30-60/candidate (https://www.imocha.io/pricing).

**Vervoe** is its own category. The pitch is that multiple-choice tests do not predict performance — what predicts performance is watching someone do the actual work. Vervoe candidates record video, write essays, build mock spreadsheets, debug code, and respond to customer scenarios. The AI grades the open-ended output against rubrics the platform learns from your highest-performing hires. Pricing starts at $300/month for Standard (10 hires/year) and $1,500/month for Pro (100 hires/year) per https://vervoe.com/pricing/. The 'per hire' billing unit is the tell — Vervoe priced itself for selection, not screening.

**Plum** is psychometric talent science wearing a SaaS skin. The product measures personality, cognitive ability, and social intelligence using established I/O psychology instruments, then matches candidates against a 'Plum Profile' built for the role. There is no coding sandbox and no work sample. You buy Plum when you are hiring for leadership, sales, customer success, or any role where personality and learning agility predict outcomes better than typing speed. Plum runs roughly $5K-25K/year depending on volume per https://www.plum.io/.

The cleanest mental model: HackerRank and Codility test what you can build. TestGorilla and iMocha test what you know. Vervoe tests how you would do the job. Plum tests who you are and how you think. Trying to make one tool do two of those jobs is how procurement spends $40K/year and still complains the platform 'does not predict performance.' It does — for the job it was built for.

Sticker price hides the truth. The honest comparison is cost per assessed candidate, because that is the line you will defend to finance. **HackerRank** Starter at ~$100/month with ~50 assessments works out to roughly $2/candidate at volume; Pro at ~$250/month with 200 assessments is about $1.25/candidate. Realistically most teams use less than their cap and pay closer to $4-8/candidate effective. Source: https://www.hackerrank.com/products/main/plans/.

**Codility** Starter at $169/month and Pro at $399/month do not publish assessment caps publicly, but sales-quoted volumes land around 50 and 250 respectively. That puts Codility at roughly $3.40/candidate at Starter cap, $1.60/candidate at Pro cap. Verify volumes during your sales call, because Codility ties assessment count to seat count more than HackerRank does, and small teams with one recruiter often hit the seat cap before the assessment cap (https://www.codility.com/pricing/).

**TestGorilla** is the math anomaly. Starter at $33/month includes unlimited assessments — yes, unlimited — with up to five tests per assessment. Pro at $75/month removes the test cap. Premium at $115/month adds anti-cheat hardening and SSO. If you send 200 candidates a month through Pro, you are at $0.38/candidate. That is not a typo. TestGorilla can do this because their unit economics rest on test-library reuse, not per-candidate compute (https://www.testgorilla.com/pricing/).

**Vervoe** flips the model. Standard at $300/month for 10 hires/year is $30/hire — but 'hire,' not 'candidate.' If your pass-through rate is 10%, that is $3/candidate. Pro at $1,500/month for 100 hires/year is $15/hire, roughly $1.50/candidate at the same pass rate. The catch: those hire caps are hard limits, and overages renegotiate the contract (https://vervoe.com/pricing/). **iMocha** is honest about per-candidate pricing — $30-60/candidate depending on volume tier and test depth (https://www.imocha.io/pricing). That is 10-20x TestGorilla but includes proctoring and a richer technical catalog.

**Plum** at $5K-25K/year is usually sold per-seat-of-hiring-manager or per-role-profile rather than per-candidate. At $12K/year with 500 candidates assessed, that is $24/candidate; with 5,000 candidates it is $2.40/candidate. Plum's value scales with volume and only justifies itself when you are running structured pipelines, not ad-hoc requisitions (https://www.plum.io/). Always recompute these numbers against your real funnel — pulling six months of ATS data before the sales call will save you 30% on every line item.

Every vendor in this comparison ships an 'AI' badge somewhere. Some of it is real, some is marketing wrapper around feature work that predates the LLM era. **HackerRank**'s most defensible AI is plagiarism and cheat detection — webcam analysis, IP fingerprinting, behavioral signals during a coding session, and similarity scoring against a corpus of past submissions. They have also shipped AI question generation in 2025 for tech leads who want custom problem variants. Useful, but not transformative.

**Codility**'s AI story is similar but quieter. They publish a similarity score and an 'AI assistance detected' flag that fires when a candidate's code looks like Copilot or ChatGPT output (https://www.codility.com/). Both Codility and HackerRank are now in the awkward spot where their core product — 'did this candidate write this code?' — got harder because every dev has Copilot. Their pricing has not dropped in response.

**TestGorilla**'s AI is largely proctoring (face detection, multi-person detection, browser-tab leaving, IP changes) and adaptive question selection inside multi-test assessments. Honest, low-glam, works. They have started rolling out AI skill simulators in 2026 — small interactive scenarios — but the core library is still item-bank based (https://www.testgorilla.com/).

**Vervoe** is the only platform here where AI is the product, not a wrapper. The AI grades open-ended text responses, scores video answers on content (not appearance), and runs rubric-based scoring on work-sample submissions. The model trains on your past hires — high performers anchor 'good,' low performers anchor 'not good,' and the rubric improves. That is genuinely useful when you are hiring 200 customer-success reps and cannot watch 200 videos. It is also where Vervoe earns the $1,500/month Pro price.

**iMocha**'s AI plays in two places: AI proctoring (similar to TestGorilla but more aggressive) and AI-authored test items at the Enterprise tier, where the platform will generate custom assessments from a job description. **Plum**'s AI is the psychometric matching algorithm itself — a multi-trait predictive model trained on years of performance outcome data. Plum's defensibility is the data, not the algorithm. If you want a buyer's frame: HackerRank/Codility AI is for catching cheaters, TestGorilla/iMocha AI is for scaling proctoring, Vervoe AI is for replacing human graders, and Plum AI is for predicting outcomes. Different products, different value, different prices.

An assessment tool that does not pipe results into your ATS will be abandoned within 90 days, no matter how good the tests are. **HackerRank** has the deepest integration matrix — Greenhouse, Lever, Workday, SmartRecruiters, iCIMS, BambooHR, JazzHR, and 30+ more, plus a documented public API. Triggering an assessment from a Greenhouse stage transition is one-click setup, and scorecards land back in Greenhouse as attachments with deep links. This is the table-stakes baseline for a Pro-tier engineering hiring tool.

**Codility** matches HackerRank on the major ATS connectors (Greenhouse, Lever, Workday, SuccessFactors, SmartRecruiters) but is thinner on the long tail. If your ATS is a Workday subsidiary or a regional European platform, Codility's EU-first posture often gives you a better integration story than HackerRank's US-centric one (https://www.codility.com/integrations/).

**TestGorilla** integrates with Greenhouse, Lever, Workable, BambooHR, JazzHR, and 20+ more via API, but the workflow is one notch less polished — results land in the ATS, but stage-triggered assessments often require Zapier or a webhook hop. Acceptable for SMBs, occasionally painful for enterprises with strict workflow audits (https://www.testgorilla.com/integrations/).

**Vervoe** has Greenhouse, Lever, JazzHR, Workable, and SmartRecruiters connectors. Because Vervoe assessments are longer (15-45 minutes of work samples), the integration also has to handle async candidate experience — reminders, deadline tracking, partial-submission states — which Vervoe handles natively (https://vervoe.com/integrations). **iMocha** plays at the enterprise tier with Workday, SuccessFactors, Taleo, iCIMS, plus Greenhouse and Lever; the per-candidate billing model makes ATS-triggered assessments straightforward to meter.

**Plum** integrates with Greenhouse, Lever, Workday, iCIMS, and exposes a clean API. Because Plum runs once per candidate (not per role), the integration pattern is different — you assess a candidate, store the Plum Profile, and reuse it across requisitions. That changes the data model your TA ops team has to maintain, and it is something to ask Plum to demo against your ATS specifically before signing (https://www.plum.io/integrations).

List price is the starting point of a conversation, not the end. **HackerRank** Starter is $100/month list, but volume buyers (50+ engineers hired/year) routinely land Pro for $200/month or get Enterprise with 1,000+ assessments and SAML for $25K-50K/year. Annual prepay yields a ~17% discount. Procurement leverage: HackerRank is publicly traded competitor pressure from Codility means salespeople will move on annual contracts > $20K (https://www.hackerrank.com/products/main/plans/).

**Codility** Starter $169 and Pro $399 are list per https://www.codility.com/pricing/. Real-world Enterprise contracts come in at $15K-60K/year depending on seat count and assessment volume. Codility's sales team has historically been more willing to flex on annual prepay (~20%) and bundle CodeLive seats than HackerRank. If you have a multi-year Codility incumbency, mention it; if you are switching from HackerRank, mention that too.

**TestGorilla**'s published pricing is closer to the truth than any other vendor here. $33, $75, $115/month are real and there is a free plan. Enterprise quotes appear above 500 candidates/month with custom SSO and audit needs, generally landing $5K-20K/year. There is little procurement leverage because the list price is already aggressive (https://www.testgorilla.com/pricing/).

**Vervoe** publishes $300 Standard and $1,500 Pro tiers, but the Pro tier with annual commit is the real entry point for any team hiring 50+ people/year. Enterprise pricing is custom, frequently $30K-80K/year, and Vervoe will discount aggressively if you commit to a multi-year deal because their gross margins depend on long contracts (https://vervoe.com/pricing/).

**iMocha** does not publish list pricing — the $30-60/candidate range comes from buyer interviews and reseller quotes. Enterprise contracts typically land $40K-150K/year for 1,000-5,000 candidates with private cloud, custom test authoring, and dedicated CSM (https://www.imocha.io/pricing). **Plum** is the most opaque — quotes vary 3-4x based on company size and use case. Expect $5K for small-business, $25K-50K for mid-market, $100K+ for enterprise leadership-pipeline programs. As of June 2026 — verify at https://www.plum.io/ before signing, because Plum repriced twice in the last 18 months.

Pure engineering hiring, 100+ devs/year, US-based, Greenhouse stack: **HackerRank** Pro or Enterprise. CodePair plus AI cheat detection plus the deepest ATS matrix wins, and at this volume you will negotiate Enterprise pricing that lands competitively per https://www.hackerrank.com/products/main/plans/. Codility is the credible alternative if you have hated HackerRank's task quality before — Codility's problems are more rigorously calibrated and their false-positive rate on plagiarism is lower in our experience.

European engineering hiring with GDPR sensitivity, Workday or SAP SuccessFactors stack: **Codility**. EU data residency is native, the company is Polish, and their Enterprise team understands DPA negotiations without a $50K legal escalation. HackerRank can serve EU but Codility leans in (https://www.codility.com/).

SMB hiring across mixed roles — engineers, sales, ops, customer success — under $1K/month budget: **TestGorilla** Pro at $75/month. The 400+ test library covers 80% of role-specific needs, anti-cheat is solid, and the unit economics let you over-assess. If you grow past 500 candidates/month and need SSO, jump to Premium at $115/month (https://www.testgorilla.com/pricing/). Genuine no-brainer at this scale.

High-volume non-technical hiring — customer support, retail, sales — where you need to test how someone actually responds to a customer, not whether they can list product features: **Vervoe**. The AI-graded video and work-sample model earns its premium when you are evaluating 1,000+ candidates for a 50-seat support center. At $1,500/month Pro it is roughly $15/hire — cheaper than a 30-minute human screen (https://vervoe.com/pricing/).

Enterprise with 3,000+ candidates/year across diverse skill areas needing per-candidate cost predictability and private cloud: **iMocha**. The catalog depth (3,000+ tests) and private-cloud option win over Workday/SuccessFactors enterprise buyers (https://www.imocha.io/pricing). Leadership pipelines, exec search, and any role where culture-fit and learning agility matter more than current skills: **Plum**. You are not testing Excel — you are predicting whether this VP will still be here in three years and outperforming peers (https://www.plum.io/). Different question, different tool.

Every coding-assessment vendor has the same 2025-2026 problem: candidates have Copilot, ChatGPT, and Claude open in another window, and increasingly on another laptop pointed at the same screen. AI-generated code passes most test suites the first time. **HackerRank** answered with stricter proctoring — full-screen lock, webcam recording, tab-leave detection, IP fingerprinting — plus an AI similarity score that flags code matching LLM output patterns. It is good, not perfect. Sophisticated candidates still bypass it.

**Codility** added a similar AI-assistance detector and an option to run a CodeLive session — a recorded pair-programming round where you watch the candidate type. CodeLive is the only honest answer right now: if you need to know whether they can code, watch them code. Both vendors have started recommending CodeLive/CodePair as the source-of-truth round and treating async assessments as a filter, not a decision (https://www.codility.com/, https://www.hackerrank.com/).

**TestGorilla** leans hard into proctoring: webcam snapshots throughout, IP and device fingerprinting, multi-person detection, browser-tab leaving, full-screen requirement. For non-coding tests (cognitive, personality, role knowledge) this is enough. For coding tests, the same LLM-bypass problem applies. TestGorilla's anti-cheat at $115/month Premium is the most affordable serious proctoring in the market (https://www.testgorilla.com/pricing/).

**Vervoe**'s answer is structural — when the test is 'record a 90-second video answering this scenario' or 'build this spreadsheet from this CSV,' ChatGPT cannot fake the video and the spreadsheet output reveals the candidate's actual approach. Open-response and work-sample testing is naturally more LLM-resistant, which is part of why Vervoe is winning brand recognition in 2026 even at premium pricing (https://vervoe.com/).

**iMocha** ships aggressive AI proctoring plus optional live-proctor human review at higher tiers, used heavily by enterprises that need audit-defensible hiring records (financial services, government, regulated healthcare). **Plum** does not have this problem because the instruments are psychometric — there is no 'right answer' to fake in a personality assessment, and Plum's algorithms detect socially-desirable response patterns automatically. If your hiring program is being audited by EEOC or by a customer's vendor-security review, ask each vendor for their adverse-impact study and validity documentation — TestGorilla, Vervoe, iMocha, and Plum all publish them; HackerRank and Codility are catching up.

The line item on the invoice is not the total cost. **HackerRank** Pro at $250/month is real, but if you need a custom problem library mapped to your engineering levels, plan 20-40 hours of engineering manager time for calibration. HackerRank will sell you a calibration package (~$5K-15K) at Enterprise. Most teams skip it and recalibrate over the first year.

**Codility** is similar — strong out-of-the-box library, but mapping tasks to your specific role rubrics takes work. Codility's customer success team is sharper than HackerRank's at calibration support, which matters more than the $100/month tier difference (https://www.codility.com/).

**TestGorilla** is fastest to deploy — pick five tests, write a job description, send a link. The hidden cost is that the library tests are generic; if you need role-specific behavioral or technical scenarios, you will either author them yourself in the platform or accept that the test is a filter, not a final decision (https://www.testgorilla.com/).

**Vervoe** has the highest implementation cost in the comparison because the AI grading rubrics improve with every hire you score. Plan a 60-90 day calibration period where you score the first batch of assessments manually so the AI learns your standards. After that the platform pays for itself in graders' time, but the first quarter is a real lift (https://vervoe.com/).

**iMocha** Enterprise typically includes 30-60 days of CSM-led implementation, custom assessment authoring, and ATS integration tuning — bundled into the $40K+ contracts. **Plum** requires building a Plum Profile per role (a workshop with hiring managers and incumbent top performers) which the CSM facilitates; budget 2-4 hours per role profile. Across all six vendors, the durable rule is that the cheaper the sticker, the more internal time you spend calibrating; the more expensive the contract, the more the vendor does for you. Price that trade-off into the procurement decision honestly.

Seed-to-Series-A startup hiring 10-30 people/year across mixed roles, budget under $200/month: **TestGorilla** Starter or Pro. Nothing else makes economic sense at this stage, and the 400+ test library covers most of what you need (https://www.testgorilla.com/pricing/). Skip the engineering-only platforms until you are hiring 25+ devs/year.

Series B-C engineering-heavy company hiring 50-150 devs/year: **HackerRank** Pro or **Codility** Pro, depending on geography. Both will get you past the SMB ceiling on assessment quality and proctoring. Run a 30-day side-by-side trial on the same role and let the engineering managers vote; the loser will not be far behind, but the winner will be obvious within two weeks.

Mid-market company hiring 200+ across diverse roles, especially with significant non-technical volume (customer success, sales, support, ops): a **Vervoe** + **TestGorilla** stack. Vervoe Pro for the work-sample assessment on high-volume roles where 'how would you handle this customer' matters more than 'do you know Excel,' TestGorilla for cognitive screens and role-knowledge gates on everything else. Total cost lands around $1,500-2,000/month, which is less than a single recruiter coordinator.

Enterprise (1,000+ employees, complex compliance, multiple hiring regions): **HackerRank** or **Codility** Enterprise for engineering, **iMocha** for high-volume regulated roles needing private cloud and audit logs, **Plum** for leadership pipelines and executive search. Total annual spend $80K-300K depending on volume. Negotiate hard at renewal — every one of these vendors has competition now and salespeople know it.

If you take only one thing from this comparison: do not buy a coding-interview platform for non-coding hiring, do not buy a personality test for software engineer hiring, and do not pay enterprise prices for SMB volume. The right tool at the wrong scale is the wrong tool. Verify all pricing at the vendor URL — as of June 2026 — verify at vendor.com/pricing before procurement, because every vendor in this comparison adjusted list price at least once in the last 12 months.