Candidate Matching API Comparison 2026: 8 APIs Compared

A candidate matching API scores how well a resume fits a job. The main choice is whether you want to send a resume and a job description and get a score back right away (HireLayer, SharpAPI), or index candidates first and then search and match inside that index (Textkernel, RChilli, Affinda, HrFlow.ai). The second also needs explanations: only some APIs say why a candidate scored the way they did. This guide compares eight matching APIs on how they work, what they explain, what a match costs and how you get access.

HireLayer is our product and it is in the list. Every other fact comes from the vendor's own documentation or pricing page, checked on 7 October 2026; the links are at the end. When a vendor does not publish something, we say so rather than guess.

The short list

API	Index first?	Explains the score	Published price
HireLayer Match and Rank	No	Status and evidence per criterion	1 credit per call, from €0.0445; Rank covers up to 10 candidates per call
Textkernel Search & Match	Yes	Which query parts matched	0.05 credit per match, 1 credit to index a document; plans from $99 a month
RChilli Search & Match	Yes	Score per entity, with weights you set	2 credits per oneMatch or index; 500 credits for $75 a month
Affinda Search & Match	Parsed documents	Score per criterion	Not published, through sales
HrFlow.ai Scoring	Yes	A probability; explanations are a separate API	Not published
SharpAPI Job Match Score	No	Scores and explanations per criterion	Word-based plans from $50 a month
JANZZsme!	Not published	Not published	From $1,500 a month (start-up licence example)
Hirize AI Matcher	No	Response not documented	2 credits per search + 1 per resume; credit price in a calculator

Two ways matching APIs work

Stateless matching takes a job and one or more resumes in the request and returns scores in the response. Nothing is stored between calls, so it fits screening the applicants of one job, ranking a shortlist, or adding a fit score inside an ATS when a candidate applies. You keep the data in your own database.

Index-based matching asks you to parse and index candidates and jobs first. Matching then runs against the whole index: "find the best candidates in my database for this new job". It is the right model for a staffing agency searching hundreds of thousands of past applicants, but it means the vendor stores your candidate data and you pay to index every document, even the ones that are never matched.

Some vendors offer both. Before you compare prices, decide wich model you need: a cheap match query on an index can still cost more than a stateless call once you add indexing.

The APIs in detail

HireLayer Match and Rank

HireLayer splits matching into small steps. Job Extract turns a job description into weighted criteria (weight 1 to 3, a mandatory flag and a rationale). Match then evaluates one resume against those criteria and returns, for each one, a status (ideal, potential, not_mentioned or not_valid) and the evidence from the resume, plus a weighted score from 0 to 1 and a summary. Rank takes one job description and up to 10 resumes and returns them in order, each with a score and a rationale.

Everything is stateless and every call costs one credit, so one Rank call ranks up to 10 candidates for the price of one. Credits cost from €0.0445 each and the free plan includes 50 a month. One limit to know: in the current version, criteria labels, explanations and rationales are written in French, whatever the language of the resume.

Best for: screening the applicants of a job with an explanation a recruiter can check, without storing candidates at a vendor.

Textkernel Search & Match

Textkernel (now part of Bullhorn, and formerly Sovren) offers Search & Match V2 on its Tx Platform. You index candidate and job documents, then match a source document against the index; each result has a score and the query parts it matched. The docs describe it as a term-based engine. A match query costs 0.05 credit, indexing a document costs 1 credit, and parsing plus indexing 2 to 2.3 credits. Professional plans start at $99 a month and the trial includes 500 free credits. Hosting is available in the US, EU and Australia. Its older Bimetric scoring endpoint is marked deprecated in the docs.

Best for: searching a large candidate database with transparent, keyword-level matching.

RChilli Search & Match

RChilli also matches inside an index: documents must be indexed before you search or match them. The match response includes an explainScore object with a score and a maximum score for each entity, such as job profile, required skills or years of experience, and you can configure the weights. Indexing and oneMatch cost 2 credits each, and search calls cost nothing extra. Credits start at $75 a month for 500, with 100 free credits to try it. Servers are in the US, the EU or Singapore.

Best for: teams already using RChilli parsing who want configurable, entity-level scores.

Affinda Search & Match

Affinda's match endpoint takes a parsed resume and a parsed job description and returns a score from 0 to 1 with a sub-score per criterion: job title, management level, experience, skills, languages, location, education and occupation group. It is documented in the platform API, but the pricing page excludes job description parsing and Search & Match, which are sold through the sales team.

Best for: Affinda customers who want a criterion-level breakdown on parsed documents.

HrFlow.ai Scoring and Grading

HrFlow.ai scores profiles indexed in a Source against a job indexed in a Board. The Scoring endpoint ranks many profiles for one job and Grading scores up to 32 profiles one to one; both return a probability per profile, with no breakdown. HrFlow presents explanations as a seperate Reasoning API. It claims more than 43 languages and cross-lingual ranking. Prices are not published: the pricing page returned an error when we checked.

Best for: teams that want a hosted HR data layer where profiles and jobs live with the vendor.

SharpAPI

SharpAPI's Resume/CV Job Match Score takes a resume file and the job description text in one request and returns scores per criterion (skills, education, experience, certifications, soft skills and more) with readable explanations, in a language you can choose. Calls are asynchronous. Pricing is based on words processed, from $50 a month for 250,000 words, with a 14-day trial of 100,000 words.

Best for: developers who want explained scores straight from a file, billed by volume of text.

JANZZ, Hirize and DaXtra

JANZZ.technology sells concept-based skills and job matching (JANZZsme!) on 1:1 or 1:n basis, with a start-up licence example at $1,500 a month for up to 10,000 transactions and trials on request. Hirize documents an AI Matcher endpoint that takes a job title, seniority and skills plus a resume; it costs 2 credits per search plus 1 per resume, but the response format and the price per credit are not published in text. DaXtra markets ranking and scoring "as an embeddable API" with a sample response, but its public reference docs only cover parsing, so access goes through a demo.

How to choose a candidate matching API

Pick the model first. Screening applicants for one job is a stateless job. Searching your whole database is an index job.
Ask for the why, not just the score. A recruiter has to defend a shortlist. Criterion-level statuses or sub-scores let them check the evidence; a single probability does not.
Count the full cost. Add indexing, parsing and the number of candidates per call. One Rank call for 10 candidates is not the same price as ten match calls.
Check languages and regions. Both the resume language and the language of the explanations matter, and so does where candidate data is processed and stored.
Plan for human oversight. Under the EU AI Act, systems that evaluate or rank candidates are high-risk. Keep a person in charge of the decision and log what the API returned. Our EU AI Act recruitment checklist goes through the requirments.

To test the stateless approach on your own job, the HireLayer Match API and the HireLayer Rank API pages show full request and response examples, and the screening pipeline guide chains parsing, criteria, ranking and matching end to end.

Try the live demo Read the API documentation View pricing

Frequently asked questions

What is a candidate matching API?

An API that compares a candidate with a job and returns a score, and sometimes an explanation. Some take the resume and job in the request; others match inside a database of candidates you indexed first.

Which API can rank several candidates for one job?

HireLayer Rank orders up to 10 candidates for one job description in a single call, with a score and a rationale for each. Textkernel, RChilli and HrFlow.ai rank candidates from an index you build first.

Which matching APIs explain their scores?

HireLayer returns a status and evidence per criterion, Affinda and RChilli return sub-scores per criterion or entity, SharpAPI returns scores with explanations, and Textkernel shows which query parts matched. HrFlow.ai's scoring returns a probability, with explanations offered by a separate API.

How much does candidate matching cost?

It depends on the model. HireLayer charges 1 credit per Match or Rank call (from €0.0445). Textkernel charges 0.05 credit per match plus 1 credit per indexed document, and RChilli 2 credits per oneMatch or indexed document. Affinda, HrFlow.ai and DaXtra do not publish matching prices.

Can I use an LLM directly instead of a matching API?

You can, but you then own the prompt, the scoring scale, the consistency between calls and the audit trail. A matching API gives you a fixed response format and a documented scale, which makes scores comparable across candidates and easier to explain.

Sources and further reading

Louis Desclous

Published on October 7, 2026 · Reading time: 9 minutes