Skip to content

Reducto · Document API

Reducto Deep Extract on ResumeBench: 85.6 F1, rank 2 of 18

Reducto Deep Extract ranks 2nd of 18 on ResumeBench: 85.6 F1, $40 per 1,000 pages, 65 seconds per resume. It is 1.6 points behind LlamaParse Extract Agentic Plus.

2nd of 7 document APIs. Strongest on scanned & rotated (+7.2 points), rotated & warped pages (+6.3 points) and hidden text & prompt injections (+5.5 points). Weakest on resume-builder templates (−3.6 points), handwritten (−2.1 points) and table grid (−0.9 points).

Rank
2/18
F1
85.6
Precision
87.5
Recall
87.3
Perfect resumes
36/300
Per 1,000 pages
$40
Median per resume
65 s

Averaged over 300 resumes. Dates (scored separately): 97.7. 90% finish within 92 s. Cost from list price.

Where Reducto Deep Extract wins and loses

F1 by kind of resume, against the median system.

Reading the page

Reducto Deep Extract by reading the page
Kind of resume# of resumesF1MedianGapRank
Handwritten949.451.5−2.116th
Image only2876.471.4+5.15th
Sidebar or grid8285.281.9+3.33rd
Photo4586.484.0+2.32nd

Distractors

Reducto Deep Extract by distractors
Kind of resume# of resumesF1MedianGapRank
Rotated & warped pages2189.382.9+6.35th
Hidden text & prompt injections1490.184.5+5.55th
Content as images888.088.2−0.211th
Wording traps1684.284.5−0.310th
Typos kept on purpose3787.983.7+4.25th

Extraction challenges

Reducto Deep Extract by extraction challenges
Kind of resume# of resumesF1MedianGapRank
Many entries8384.381.9+2.54th
Long bullet lists10184.581.4+3.14th
Split across pages8584.182.2+1.96th
Big skills list6085.883.6+2.24th
Custom sections10582.680.8+1.74th
Combined headings3782.782.7+0.09th
Promotions8886.083.5+2.53rd
Partial dates13885.283.4+1.74th
Overlapping jobs6985.981.7+4.12nd
Accented text1385.585.5−0.110th

Noise

Reducto Deep Extract by noise
Kind of resume# of resumesF1MedianGapRank
Repeated header6787.385.2+2.13rd
References5590.285.2+5.03rd
Busy page1883.680.1+3.52nd

Source

Reducto Deep Extract by source
Kind of resume# of resumesF1MedianGapRank
Everyday resumes21287.285.1+2.13rd
Expert-built (Canva / LaTeX)3284.683.6+1.06th
Resume-builder templates1077.681.2−3.615th
Scanned & rotated1990.182.9+7.25th
Distractor layouts1885.685.3+0.28th

Layout

Reducto Deep Extract by layout
Kind of resume# of resumesF1MedianGapRank
Single column19585.584.0+1.64th
Two-column2386.986.8+0.19th
Two-column sidebar5483.880.0+3.83rd
Three-column2086.084.8+1.23rd
Table grid893.294.1−0.912th

Length

Reducto Deep Extract by length
Kind of resume# of resumesF1MedianGapRank
Short14888.787.1+1.63rd
Medium9085.984.7+1.22nd
Long6277.574.6+2.86th

By section of the resume

F1 by resume section, against the median system.

Reducto Deep Extract by resume section
SectionResumesF1MedianGap
Contact details30094.293.5+0.7
Experience30090.686.3+4.3
Education29792.089.4+2.6
Summary29085.281.7+3.4
Certifications25281.481.1+0.3
Awards9479.475.4+4.0
Skills28593.992.1+1.8
Languages16781.884.3−2.6
Custom sections21251.349.9+1.4
Interests6493.891.7+2.0
Volunteering7789.584.5+5.0
Profiles3469.666.7+2.9
Projects7788.183.2+4.9
Publications5079.378.6+0.8

On the published resumes

Best and worst of the 20 public resumes.

Reducto Deep Extract on published resumes
ResumeWhat makes it hardF1
Senior Manufacturing EngineerThird-person prose, no bullets93.9
Deputy Head of MissionRotated scan with hidden identity text89.2
Lead Respiratory Therapist (Adult ICU)Long resume, 34 entries88.9
senior Consultant - package ImplementationHandwritten resume54.8
Senior Game DesignerGame designer's Canva resume52.9
Senior ocularist and practice ownerHandwritten resume, six pages37.4

How it was run

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

Reducto Deep Extract on Reducto’s site

Run the open-source harness: list its pipelines, or grade your own predictions (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 providers
$ resume-bench grade-file my_predictions.jsonl --split test

Closest alternatives

Nearest in F1.

Embed this result

Shows its current rank. Updates with each run.

Reducto Deep Extract: #2 on ResumeBench resume extraction benchmark
<a href="https://www.resumebench.ai/systems/reducto-deep-extract"><img src="https://www.resumebench.ai/badge/reducto-deep-extract.svg" alt="Reducto Deep Extract: #2 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 Deep Extract at resume extraction?

85.6 F1 on ResumeBench's 300 resumes, 2nd of 18 models.

How much does Reducto Deep Extract cost to parse resumes?

$40 per 1,000 pages, about $0.112 per resume.

How fast is Reducto Deep Extract?

65 seconds per resume (median).

Which resumes is Reducto Deep Extract best at?

Scanned & rotated, rotated & warped pages and hidden text & prompt injections. Weakest on resume-builder templates, handwritten and table grid.

What is a cheaper alternative to Reducto Deep Extract?

LlamaParse Extract Agentic: 85.5 F1 for $23 per 1,000 pages.