Skip to content

Reducto · Document API

Reducto Extract on ResumeBench: 81.7 F1, rank 12 of 18

Reducto Extract ranks 12th of 18 on ResumeBench: 81.7 F1, $20 per 1,000 pages, 16 seconds per resume. It is 5.5 points behind LlamaParse Extract Agentic Plus. Qwen 3.8 Flash scores 84.1 for $2.98.

6th of 7 document APIs. Strongest on scanned & rotated (+7.0 points), rotated & warped pages (+6.3 points) and hidden text & prompt injections (+2.7 points). Weakest on handwritten (−12.1 points), three-column (−4.9 points) and busy page (−4.7 points).

Rank
12/18
F1
81.7
Precision
83.7
Recall
84.0
Perfect resumes
16/300
Per 1,000 pages
$20
Median per resume
16 s

Averaged over 300 resumes. Dates (scored separately): 96.8. 90% finish within 633 s. Cost from list price.

Where Reducto Extract wins and loses

F1 by kind of resume, against the median system.

Reading the page

Reducto Extract by reading the page
Kind of resume# of resumesF1MedianGapRank
Handwritten939.451.5−12.118th
Image only2872.071.4+0.69th
Sidebar or grid8282.181.9+0.29th
Photo4581.984.0−2.112th

Distractors

Reducto Extract by distractors
Kind of resume# of resumesF1MedianGapRank
Rotated & warped pages2189.282.9+6.36th
Hidden text & prompt injections1487.284.5+2.78th
Content as images884.988.2−3.313th
Wording traps1681.484.5−3.113th
Typos kept on purpose3786.283.7+2.66th

Extraction challenges

Reducto Extract by extraction challenges
Kind of resume# of resumesF1MedianGapRank
Many entries8379.581.9−2.413th
Long bullet lists10180.281.4−1.212th
Split across pages8579.782.2−2.512th
Big skills list6083.283.6−0.411th
Custom sections10578.280.8−2.612th
Combined headings3780.282.7−2.512th
Promotions8882.883.5−0.712th
Partial dates13880.083.4−3.413th
Overlapping jobs6980.481.7−1.412th
Accented text1387.885.5+2.24th

Noise

Reducto Extract by noise
Kind of resume# of resumesF1MedianGapRank
Repeated header6783.685.2−1.612th
References5584.485.2−0.812th
Busy page1875.480.1−4.713th

Source

Reducto Extract by source
Kind of resume# of resumesF1MedianGapRank
Everyday resumes21283.085.1−2.212th
Expert-built (Canva / LaTeX)3280.083.6−3.613th
Resume-builder templates1081.881.2+0.69th
Scanned & rotated1989.982.9+7.06th
Distractor layouts1882.385.3−3.113th

Layout

Reducto Extract by layout
Kind of resume# of resumesF1MedianGapRank
Single column19581.384.0−2.613th
Two-column2383.286.8−3.612th
Two-column sidebar5481.680.0+1.68th
Three-column2079.984.8−4.913th
Table grid890.894.1−3.314th

Length

Reducto Extract by length
Kind of resume# of resumesF1MedianGapRank
Short14885.387.1−1.811th
Medium9082.584.7−2.212th
Long6271.974.6−2.813th

By section of the resume

F1 by resume section, against the median system.

Reducto Extract by resume section
SectionResumesF1MedianGap
Contact details30093.493.5−0.1
Experience30084.086.3−2.3
Education29782.389.4−7.1
Summary29081.781.7+0.0
Certifications24481.381.1+0.2
Awards10273.575.4−1.9
Skills29093.092.1+0.9
Languages16984.384.3+0.0
Custom sections21644.449.9−5.5
Interests6492.891.7+1.1
Volunteering8082.784.5−1.8
Profiles3558.166.7−8.6
Projects7884.683.2+1.4
Publications5176.078.6−2.5

On the published resumes

Best and worst of the 20 public resumes.

Reducto Extract on published resumes
ResumeWhat makes it hardF1
Head of Developer Experience (Internal title: Jedi Master of Analytics) Unusual job titles93.5
Senior Manufacturing EngineerThird-person prose, no bullets92.4
Deputy Head of MissionRotated scan with hidden identity text89.3
Senior Game DesignerGame designer's Canva resume61.7
Senior ocularist and practice ownerHandwritten resume, six pages47.0
senior Consultant - package ImplementationHandwritten resume40.0

How it was run

Reducto Extract API, default settings. Same PDF, schema and instructions as every other system, one run per resume (4 November 2026).

Reducto Extract on Reducto’s site

Reproduce it with the open-source harness (GitHub):

$ pip install "resume-bench[runners] @ git+https://github.com/careerflow/ResumeBench"
$ hf auth login   # the dataset is gated: accept its terms on Hugging Face first
$ resume-bench download
$ resume-bench run reducto_extract --split test
$ resume-bench grade reducto_extract --split test

Closest alternatives

Nearest in F1.

Embed this result

Shows its current rank. Updates with each run.

Reducto Extract: #12 on ResumeBench resume extraction benchmark
<a href="https://www.resumebench.ai/systems/reducto-extract"><img src="https://www.resumebench.ai/badge/reducto-extract.svg" alt="Reducto Extract: #12 on ResumeBench resume extraction benchmark" height="28"></a>

Choosing a resume extraction model

All 18 models · Leaderboard

Frequently asked questions (FAQs)

How accurate is Reducto Extract at resume extraction?

81.7 F1 on ResumeBench's 300 resumes, 12th of 18 models.

How much does Reducto Extract cost to parse resumes?

$20 per 1,000 pages, about $0.056 per resume.

How fast is Reducto Extract?

16 seconds per resume (median).

Which resumes is Reducto Extract best at?

Scanned & rotated, rotated & warped pages and hidden text & prompt injections. Weakest on handwritten, three-column and busy page.

What is a cheaper alternative to Reducto Extract?

Qwen 3.8 Flash: 84.1 F1 for $2.98 per 1,000 pages.