Skip to content

Extend · Document API

Extend Extract on ResumeBench: 81.0 F1, rank 13 of 18

Extend Extract ranks 13th of 18 on ResumeBench: 81.0 F1, $12 per 1,000 pages, 31 seconds per resume. It is 6.2 points behind LlamaParse Extract Agentic Plus. Qwen 3.8 Flash scores 84.1 for $2.98.

7th of 7 document APIs. Strongest on handwritten (+1.0 points), content as images (+0.0 points) and distractor layouts (−0.1 points). Weakest on expert-built (canva / latex) (−6.5 points), two-column (−5.5 points) and photo (−5.5 points).

Rank
13/18
F1
81.0
Precision
82.9
Recall
83.2
Perfect resumes
16/300
Per 1,000 pages
$12
Median per resume
31 s

Averaged over 300 resumes. Dates (scored separately): 96.6. 90% finish within 50 s. Cost from list price.

Where Extend Extract wins and loses

F1 by kind of resume, against the median system.

Reading the page

Extend Extract by reading the page
Kind of resume# of resumesF1MedianGapRank
Handwritten952.551.5+1.08th
Image only2870.771.4−0.610th
Sidebar or grid8280.081.9−2.013th
Photo4578.584.0−5.514th

Distractors

Extend Extract by distractors
Kind of resume# of resumesF1MedianGapRank
Rotated & warped pages2179.882.9−3.110th
Hidden text & prompt injections1482.484.5−2.110th
Content as images888.288.2+0.09th
Wording traps1682.584.5−2.011th
Typos kept on purpose3782.383.7−1.411th

Extraction challenges

Extend Extract by extraction challenges
Kind of resume# of resumesF1MedianGapRank
Many entries8380.281.9−1.611th
Long bullet lists10179.881.4−1.713th
Split across pages8579.682.2−2.613th
Big skills list6081.483.6−2.213th
Custom sections10577.380.8−3.513th
Combined headings3778.082.7−4.713th
Promotions8881.783.5−1.813th
Partial dates13880.183.4−3.412th
Overlapping jobs6981.281.7−0.511th
Accented text1384.385.5−1.312th

Noise

Extend Extract by noise
Kind of resume# of resumesF1MedianGapRank
Repeated header6782.285.2−3.013th
References5584.885.2−0.411th
Busy page1878.280.1−1.911th

Source

Extend Extract by source
Kind of resume# of resumesF1MedianGapRank
Everyday resumes21282.685.1−2.613th
Expert-built (Canva / LaTeX)3277.183.6−6.515th
Resume-builder templates1080.181.2−1.012th
Scanned & rotated1979.582.9−3.410th
Distractor layouts1885.285.3−0.110th

Layout

Extend Extract by layout
Kind of resume# of resumesF1MedianGapRank
Single column19581.484.0−2.612th
Two-column2381.386.8−5.513th
Two-column sidebar5478.380.0−1.712th
Three-column2080.484.8−4.412th
Table grid889.894.1−4.215th

Length

Extend Extract by length
Kind of resume# of resumesF1MedianGapRank
Short14883.787.1−3.413th
Medium9081.684.7−3.113th
Long6273.674.6−1.011th

By section of the resume

F1 by resume section, against the median system.

Extend Extract by resume section
SectionResumesF1MedianGap
Contact details30091.193.5−2.5
Experience30081.086.3−5.4
Education29790.289.4+0.8
Summary29080.781.7−1.0
Certifications25475.781.1−5.4
Awards9973.875.4−1.5
Skills28788.792.1−3.4
Languages16785.184.3+0.8
Custom sections21244.449.9−5.5
Interests6490.991.7−0.8
Volunteering7784.584.5−0.1
Profiles3642.666.7−24.1
Projects8280.283.2−3.0
Publications5185.178.6+6.5

On the published resumes

Best and worst of the 20 public resumes.

Extend Extract on published resumes
ResumeWhat makes it hardF1
Senior Manufacturing EngineerThird-person prose, no bullets94.5
Lead Respiratory Therapist (Adult ICU)Long resume, 34 entries88.3
Head of Events & Public Space OperationsTable grid with emoji bullets88.3
Head of Developer Experience (Internal title: Jedi Master of Analytics) Unusual job titles55.1
Postdoctoral ResearcherAcademic CV, seven pages54.2
Senior ocularist and practice ownerHandwritten resume, six pages44.0

How it was run

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

Extend Extract on Extend’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.

Extend Extract: #13 on ResumeBench resume extraction benchmark
<a href="https://www.resumebench.ai/systems/extend-extract"><img src="https://www.resumebench.ai/badge/extend-extract.svg" alt="Extend Extract: #13 on ResumeBench resume extraction benchmark" height="28"></a>

Choosing a resume extraction model

All 18 models · Leaderboard

Frequently asked questions (FAQs)

How accurate is Extend Extract at resume extraction?

81.0 F1 on ResumeBench's 300 resumes, 13th of 18 models.

How much does Extend Extract cost to parse resumes?

$12 per 1,000 pages, about $0.033 per resume.

How fast is Extend Extract?

31 seconds per resume (median).

Which resumes is Extend Extract best at?

Handwritten, content as images and distractor layouts. Weakest on expert-built (canva / latex), two-column and photo.

What is a cheaper alternative to Extend Extract?

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