CodeSignal

What it is

CodeSignal sells technical assessment, but it no longer describes itself that way. The company’s own 2026 positioning is “the AI-native skills platform,” built on five product lines: AI Interviewer, Skills Assessments, Live Tech Interviews, Skills Development, and Skills Intelligence. Assessments now reach sales, finance, and marketing roles alongside engineering, and a second business — CodeSignal Learn, free for individuals with a $24.99/month Cosmo+ tier — sells upskilling to the same buyer. Founded in 2015 by Tigran Sloyan and three co-founders, CodeSignal raised a $50M Series C led by Index Ventures in 2021, and Latka’s tracker put 2024 revenue near $53.8M. Named customers include Google, Meta, Anthropic, Dropbox, and Ramp.

Two changes matter more than the product-line count. The Hire suite publishes list prices now. And CodeSignal inverted its position on AI in the assessment.

Why it shows up in Recruiting stacks

• The assessment measures AI use instead of blocking it. Since May 2025 the Hire suite ships two AI-assisted modes. Full AI Co-Pilot Mode puts Cosmo, CodeSignal’s in-IDE assistant, into the assessment as an active collaborator; Guided Support Mode restricts it to navigation, syntax, and documentation lookup. Hiring teams get the full candidate-AI transcript plus session replay, so the scored artifact is how the candidate directed the model — question quality, iteration, response to guidance — rather than whether they touched one. If your engineers ship code with a copilot every day, this is the nearest thing on the market to testing the actual job.
• AI Interviewers reach past engineering. Pre-built interviewer personas for sales, finance, marketing, and accounting run the same conversational screen against go-to-market candidates. That is a real argument for a TA team to consolidate on one vendor instead of running Apriora next to a coding tool.
• Portable scoring across test families. Certified assessments, including the General Coding Assessment, produce an Assessment Score on a 200-600 scale shared across CodeSignal’s test families, so a threshold set for one role is readable against another.
• ATS coverage is wide and native. Direct integrations for Greenhouse, Lever, Ashby, Workday, iCIMS, SmartRecruiters, and Workable, plus a self-service API key. Basic ATS connections start on Grow; enterprise ATS connections are Pro-only.

Pricing reality

CodeSignal Hire publishes list prices (checked 29 August 2026). The billing unit is a credit, and a credit burns when a candidate starts an assessment or interview — not when they pass, and not when you build the test.

• Build — $99/month, or $79/month billed annually. 5 credits/month, 60/year. Assessments, AI interviewers for product, design, and engineering, AI proctoring.
• Grow — $599/month, or $479/month billed annually. 35 credits/month, 420/year. Adds go-to-market assessments, more interviewer personas, basic ATS integrations.
• Pro — quote-only. Adds the full assessment catalog, advanced fraud prevention, enterprise ATS connections, and a dedicated CSM.
• Overage — $20 per extra credit on both published tiers, billed the following cycle.
• Learn — free for individuals, Cosmo+ at $24.99/month. Teams and Education are quote-only.

Run the per-candidate math before picking a tier. Build billed annually is $948/year for 60 credits, or $15.80 per candidate. Grow is $5,748/year for 420 credits, or $13.69. Overage at $20 sits above both. The crossover is exact: Build plus overage and Grow cost the same at 300 candidates a year. Screen fewer than 300 and Build with overage is cheaper even when you blow through the monthly allowance every cycle.

At the Pro end, Vendr’s marketplace data across 62 purchases puts the median annual contract at $24,394, in a range of $8,000 to $66,742, at an average negotiated discount near 20%.

An earlier version of this page listed no public price and called the effective entry point “mid-five figures annually.” Both were wrong. The median is low-five-figures, and the floor is now under $1,000 a year.

Annual credits do not roll into a new contract term. That is the sharpest edge in the model.

Best for

• Engineering and TA leaders screening 50-500 technical candidates a year who want the assessment to reflect copilot-assisted work rather than fight it.
• Teams already standardized on a supported ATS that want scores written back to the candidate record without middleware.
• Organizations screening engineering and go-to-market candidates who would otherwise carry two assessment vendors.

Not for you if you hire fewer than 20 engineers a year — the $948 floor is affordable, but the calibration work behind a defensible cut score is not worth it at that volume — or if your bottleneck is interview capacity rather than screening throughput. CodeSignal scores candidates; it does not supply interviewers.

Alternatives — and when to pick them instead

• HackerRank — the largest platform in the category by customer count and the default cross-shop. Pick it when your engineering org already owns a HackerRank question library it will not rebuild, or when procurement prefers one negotiated contract to a credit meter that moves with hiring volume.
• Codility — flat pricing from $100/month, positioned on algorithmic depth and task-library rigor. Pick it when you want a predictable monthly line with no overage exposure and the AI-interviewer products are not something you will turn on.
• Karat — interview-as-a-service with human interviewers, priced per interview. Pick it when the constraint is engineer hours spent interviewing, not screening throughput.
• micro1 and Apriora — the fastest-growing entrants in the adjacent AI-interviewer segment; micro1 crossed $100M ARR in December 2025, though most of that revenue comes from its AI data-training business rather than the interviewer product. Both sell the conversational AI screen without an assessment library under it. Cross-shop them directly if AI Interviewer is the only CodeSignal product you plan to use, because otherwise you are paying for a test bank you never open.

Watch-outs

• The old Coding Score is gone, and cut scores do not carry over. CodeSignal retired the 300-850 Coding Score in spring 2023 for the 200-600 Assessment Score. Guard: put your historical thresholds through CodeSignal’s published conversion before reusing them — a team that carries an 800-and-above bar into the new scale rejects every candidate it sees.
• Co-pilot mode changes what the number means. A score from an AI-assisted assessment is not comparable to one from a blocked-AI assessment, and benchmarks re-baseline with the mode. Guard: calibrate against a cohort of your own engineers taking the exact mode you plan to ship, before you set a pass mark on external candidates.
• Unused credits are forfeited, not refunded. Annual plans lock the spend, and the allowance does not carry into the next term. Guard: stay on Build and absorb overage until your hiring plan clears 300 candidates a year; the overage premium costs less than a forfeited Grow allowance.
• Advanced fraud prevention sits behind the quote. Build and Grow include AI proctoring; the deeper fraud tooling is Pro-only. Guard: if assessment fraud is the reason you are buying, get the Pro entitlement list in writing before you sign a published tier and find the gap in month two.
• The platform now spans hiring, L&D, and academic programs. The sales motion pulls toward a multi-product agreement. Guard: contract Hire on its own unless L&D holds a separate budget line — a Learn Teams attachment negotiated inside a hiring deal has no owner at renewal.