Seven Glider AI alternatives, sorted by what they actually test

Glider AI does not publish a price. I opened its pricing page on 6 September 2026 and found customer testimonials, a demo request form and a phone number, but no plans, no per-seat cost and no per-assessment cost. Jobma, its closest competitor, does the same thing. That is the first practical problem with shopping for Glider AI alternatives: a good chunk of this category hides behind a sales call, so you cannot line the options up on a spreadsheet until you have already sat through three or four demos.

So this comparison is built the other way round. Rather than ranking the tools one to seven, it sorts them by what they actually measure. That turns out to be the decision that matters, and it is the one that roundups usually skip in favour of a feature grid where every row says yes.

What Glider AI does, and what it does not

Glider AI sits in the screening layer. You send a candidate a link, they complete a coding task or a skills test or a recorded video answer, and the platform scores it and flags anything that looks like cheating. It also runs live coding interviews and does identity verification. Sensible enough, and the proctoring piece is the part hiring teams tend to buy it for.

What it does not do is help the human in the room. Once a candidate reaches a live panel, the assessment platform has finished its job and your interviewers are back to typing notes into a doc while trying to listen. Worth holding onto, because it explains why several teams end up running two tools rather than replacing one with another.

Why teams go looking for Glider AI alternatives

Three reasons come up repeatedly in this category, and they pull in different directions.

The first is price opacity. If you are a 30-person company you probably want to know whether a platform costs $400 a year or $40,000 before you book a call, and quote-only pricing filters out exactly those buyers. The second is candidate drop-off. Long proctored assessments cost you applicants, and in a market where the 2025 Stack Overflow Developer Survey found most professional developers are employed and only passively looking, an hour-long take-home is a real filter on who bothers. The third is scope. Some teams want coding tests. Others want cognitive and personality testing, or role-play simulations, and those are different products wearing similar marketing.

The seven Glider AI alternatives, and what each one measures

Prices below came from each vendor’s own pricing page on 6 September 2026. Where a vendor publishes nothing, I have said so rather than repeating a number from a review site.

HackerRank is the default answer, and mostly for a boring reason: candidates already know the interface, so you spend less of your screening budget on people fumbling the tool instead of the problem. Its Screen product page lists the ATS connector library, which is the widest in this group. If your blocker is integration rather than assessment quality, start here.

Codility reads as the closer substitute for Glider on the engineering side. Real-world tasks rather than puzzles, plagiarism detection, and CodeLive for the collaborative round. Pricing is on its own pricing page, which already puts it ahead of Glider for anyone doing a budget comparison in week one.

CodeSignal is worth a look only if you hire at enough volume for its standardised Coding Score to be useful. A portable score a candidate can reuse across employers is genuinely different from a one-off test result. At ten hires a year it is overkill.

TestGorilla goes wider than code. Cognitive ability, personality, language proficiency, situational judgement. Its pricing page publishes plans and a free tier, so a small team can pilot it in an afternoon. If you are hiring across sales, support and engineering from one budget, this is probably the shortlist entry.

Vervoe takes a different bet. Instead of asking questions, it hands candidates a simulation of the job and scores the output. I like the idea and I am not sure it scales cleanly to senior engineering roles, where the interesting part of the work is architectural judgement that a 40-minute simulation cannot really surface. For support, ops and junior sales roles it looks strong.

Mercer Mettl and Jobma are both quote-only. Mettl is the volume play, common in large India-based hiring operations, with heavy proctoring. Jobma leans on async video and AI-led interviews. Both will give you a number on a call, and neither will give you one on the web.

The proctoring question nobody answers on a demo call

Every platform in the table sells proctoring of some kind. Webcam monitoring, tab-switch detection, keystroke analysis, plagiarism scoring. It’s the feature that gets the most demo time and the least honest discussion, so here’s the version you won’t get on a sales call.

Browser-based proctoring can see what happens inside the browser. It can tell you a tab lost focus, that a paste event fired, or that a webcam feed went dark. What it can’t see is a second device, a phone propped against a monitor, or a native desktop application drawing outside the browser’s capture surface. That’s a structural limit, not a product gap, and it applies to all seven tools here in roughly the same way.

Which matters for how you read a proctoring report. A clean report doesn’t mean a candidate worked unaided. It means nothing tripped the checks that this particular tool can run. Teams that treat the score as a verdict rather than a signal end up rejecting honest candidates over a flaky webcam, which is a worse outcome than the cheating they were worried about. We’ve written up the detection side of this in more detail in our notes on what screen-share detection can and can’t see.

What five of these seven still handle badly

Assessment platforms stop at the door of the live interview. Five of the seven above (HackerRank, Codility, CodeSignal, TestGorilla and Vervoe) don’t have a product for the panel round at all, and the two video-first ones record the conversation rather than help the interviewer during it.

That gap shows up as a specific failure. An interviewer spends the first ten minutes taking notes, the middle twenty actually listening, and the last ten reconstructing what happened. Scorecards get written from memory two hours later, sometimes the next morning. If you’ve ever compared two interviewers’ notes on the same candidate and found they disagree about what the candidate actually said, this is why. It’s also, honestly, a smaller problem than a broken assessment stage, so fix the screening layer first if both are wobbly.

How to pick one in an afternoon

Pick by the constraint that is actually biting, not by feature count.

• Blocked on ATS integration: HackerRank.
• Blocked on engineering signal quality: Codility.
• Hiring non-engineering roles too: TestGorilla.
• Hiring thousands and needing proctoring at scale: Mercer Mettl, and accept the sales call.
• Screening on output rather than answers: Vervoe.
• Async video is the whole requirement: Jobma.

Then run one real requisition through it before signing anything longer than a month. Every platform here demos well.

Where Craqly fits, and where it does not

Craqly is not an assessment platform and will not replace Glider AI for screening. It works in the round the tools above hand back to you: the live interview. It runs as a desktop app, listens to the call on Zoom, Teams or Google Meet, and surfaces structured notes and follow-up prompts to the interviewer while the conversation is still happening, so the scorecard gets written from a transcript rather than from memory. The free Starter plan gives 20 credits a month, where one credit is one minute of live session, and credits reset monthly. Paid plans start at $19 a month billed yearly, or $38 month to month, checked on 6 September 2026.

If you want the candidate-facing view of this same stack, our notes on how AI resume screening actually filters applications and the questions recruiters ask and why cover the two rounds on either side of the assessment.

One prediction I could easily be wrong about: quote-only pricing in this category will not survive another two years. Buyers now compare six tools in a browser before they talk to anyone, and a page with no number on it loses that comparison silently, before anybody picks up the phone.