Affinda alternative
Affinda is a bigger product than this one and better at most of what a procurement checklist asks about. The difference worth a page is narrower and it is this: their price is a conversation, and ours is a number you can read right now.
Written by the people who build the alternative, so read it as that. Every claim about Affinda here was read from Affinda’s own pages on and is cited where it appears; none of it comes from a review site. Their pages are theirs to change, and nothing here updates when they do — so check anything you are about to rely on. Every claim about this service is checkable today against the free parser or the field reference.
What Affinda does better
This section is first because it is the one that decides most of these comparisons, and burying it would be the tell that this page was not worth reading.
| Affinda | |
|---|---|
| Third-party audits | ISO 27001:2022 certified, SOC 2 Type II, and GDPR compliant, with a public Trust Centre. |
| Data residency | Regional hosting in APAC, EU and US data centres. |
| Self-hosting | Yes — the parser runs as a container in your own infrastructure (docker run affinda/resume-parser), with 10,000 free parses on a trial licence. |
| Languages | 50+ languages. |
| Formats | PDF, DOC, DOCX, TXT, RTF, HTML, PNG, JPG, TIFF, ODT, XLS and XLSX, including scans and photographs. |
| Speed | A median parse time of 50 ms, and fifty resumes a second per container. (their own published figure) |
Read from affinda.com on 28 August 2026. Check it before you rely on it — their page is theirs to change.
The audits are the row that matters most. We hold none — no SOC 2, no ISO 27001 — and that is not a gap a description can close. Affinda publishes a Trust Centre; if your buyer asks for a report, they can produce one and we cannot. One caution on the speed row, in their favour and in ours: that is a parse time they measured, and the figure further down this page is a whole HTTP round trip measured against the live endpoint. They are not the same quantity and the difference between them is not the difference between the products.
What is different here
You can read what this costs without asking anyone. Affinda’s side of that: no figure published — the plans page offers a trial, then “Talk to us” for monthly usage and “Get a quote” for an annual commitment. That is a normal way to sell software to enterprises and it is not a criticism — but it does mean you cannot find out what Affinda costs on a Tuesday afternoon without starting a conversation with a salesperson.
Read from affinda.com on 28 August 2026. Check it before you rely on it — their page is theirs to change.
Here, the four plans are on this page, below, and the subscription is on RapidAPI. You can read the price, subscribe, and make a call inside a few minutes, without a demo, a quote, a discovery call or a procurement cycle. For a team evaluating whether resume parsing is worth building on at all, that is often the only thing standing between them and an answer.
- There is no database. Not a retention window — no store at all. The document arrives, the JSON goes back, and nothing about it outlives the response.
- No endpoint permitted to keep the document may serve a parse. Every call carries a data-retention policy excluding any model endpoint allowed to retain or train on what is sent. If no endpoint meets it the parse is refused rather than handed to one that would. That is a constant in the code with tests named after it, not a setting.
- The parser never invents. A field the CV does not state comes back
null. Nothing is inferred from a name, an area code or a school. - No score, no rank, no fit judgement. There is no such field in the schema — see whether a parser is an automated employment decision tool.
- Median parse 2.2 seconds, p90 2.4 s over 160 parses, measured end to end against the live endpoint. Fast enough to run inline on upload rather than behind a queue.
What this service does not do, including things Affinda does
Every vendor comparison is written by somebody with an interest in it, this one included. So here is the list of things this service does not do, before the part where it compares favourably at anything. If one of these rows is load-bearing for you, the honest answer is that you want a different vendor and the rest of this page will not change that.
| What | Here | When it decides the question |
|---|---|---|
| Third-party security audit | None. No SOC 2, no ISO 27001, no HIPAA attestation. | If your procurement requires an audit report, this service cannot satisfy it and no amount of description changes that. |
| Data residency | Not offered. There is no EU-only or US-only routing. | If a contract fixes where documents may be processed, buy from someone who offers a region. |
| Self-hosting | No. The parser runs only as a hosted endpoint. | If documents may not leave your network at all, a container you run yourself is the only shape that works. |
| Batch parsing | No. One document per request. | A back-catalogue of 200,000 CVs is 200,000 calls, and a vendor built for bulk will do it in fewer. |
| A contractual SLA | None published. The terms are served as is. | An uptime number you can hold somebody to is a commitment we do not make. |
| Support | An email address. There is no 24×7 desk, no ticketing portal and no account manager. | Enterprise rollouts expect a support organisation, and there is not one here. |
Stated here rather than discovered in procurement. A buyer finds out either way; finding out after an implied yes is the expensive version.
Which one fits
Choose Affinda
Your buyer requires an audit report. Documents must stay in a named region or inside your own network. You need 50+ languages, or the long tail of formats, or a latency budget far tighter than a couple of seconds. You are parsing a back-catalogue in bulk.
Choose this
You want a price you can read and a key you can have today. You need the fields a CV actually states, as typed JSON, and nothing downstream of that. Your volume is hundreds or thousands a month, not millions. Nothing in your contract requires an audit, a region, or a container you run yourself.
The plans
| Plan | Price | Included | |
|---|---|---|---|
| Basic | $0 | 25 parses a month | A hard cap, so it can never bill you. |
| Pro | $29 | 1,000 parses a month | Overage beyond the quota rather than a blocked integration. |
| Ultra | $99 | 5,000 parses a month | |
| Mega | $299 | 25,000 parses a month |
Metered and billed by RapidAPI. Basic is a hard cap and cannot bill you; the paid plans allow overage rather than blocking a working integration at the quota.
Questions
Is this a drop-in replacement for Affinda?
No, and it would be a poor page that said yes. The response shape is different, so the integration is a rewrite of the mapping layer rather than a swapped base URL. And if you are using Affinda for the audits, the residency, the self-hosted container or the language coverage, none of those exist here. The field reference is published in full so you can check the shape before writing anything.
Do you hold SOC 2 or ISO 27001?
No. This service holds no third-party security audit of any kind — no SOC 2, no ISO 27001, no HIPAA attestation. It is stated here rather than left for a procurement questionnaire to discover. What can be said instead is architectural: there is no database, so there is no store to breach, and no model endpoint permitted to retain or train on a document may serve a parse.
What does Affinda cost?
As read on their pricing page on 28 August 2026, it does not say. It offers a two-week trial with 200 credits, then "Talk to us" for monthly usage and "Get a quote" for an annual commitment. Any per-document figure you find for Affinda comes from a comparison site rather than from Affinda, and the ones published on the same afternoon disagreed with each other by a factor of six — so ask them, and treat this answer as a reading taken on one day rather than a current quote.
Is the document stored anywhere?
No. There is no database in this service — a parse is one request and nothing about it outlives the response. The document is sent to a third-party model provider to be read, which is how parsing works; every call carries a data-retention policy that excludes any endpoint permitted to keep or train on it, and a call no endpoint will take under that policy is refused rather than served by one that would.
How fast is it?
Median 2.2 seconds end to end, and a p90 of 2.4 seconds over 160 parses in three separated sittings against the live endpoint. Affinda publishes a much lower figure, and the table above quotes it — but the two are not measured the same way and subtracting one from the other would be meaningless, so this page does not. Theirs is a parse time they measured; ours is the whole HTTP round trip. Every response carries an X-Parse-Ms header, so you measure it yourself rather than trusting this page.
Can I try it without signing up?
Yes. The free parser takes a PDF, a DOCX, a photo or a scan and returns the whole response in the browser, with no account and no key. Running your own CVs through it is a better basis for a decision than any comparison table, this one included.