Skip to content

Zhipu AI · Vision LLM

GLM 5.3 Flash on ResumeBench: 59.2 F1, rank 18 of 18

GLM 5.3 Flash ranks 18th of 18 on ResumeBench: 59.2 F1, $0.80 per 1,000 pages, 46 seconds per resume. It is 27.9 points behind LlamaParse Extract Agentic Plus. Gemma 4 26B scores 62.8 for $0.56.

11th of 11 vision LLMs. Strongest on handwritten (+3.1 points), short (−7.2 points) and table grid (−11.1 points). Weakest on content as images (−77.8 points), distractor layouts (−67.5 points) and wording traps (−64.3 points).

Rank
18/18
F1
59.2
Precision
71.8
Recall
62.0
Perfect resumes
23/300
Per 1,000 pages
$0.80
Median per resume
46 s

Averaged over 300 resumes. Dates (scored separately): 63.6. 90% finish within 145 s.

Where GLM 5.3 Flash wins and loses

F1 by kind of resume, against the median system.

Reading the page

GLM 5.3 Flash by reading the page
Kind of resume# of resumesF1MedianGapRank
Handwritten954.651.5+3.14th
Image only2853.371.4−18.116th
Sidebar or grid8263.581.9−18.416th
Photo4564.084.0−20.117th

Distractors

GLM 5.3 Flash by distractors
Kind of resume# of resumesF1MedianGapRank
Rotated & warped pages2150.382.9−32.716th
Hidden text & prompt injections1446.484.5−38.117th
Content as images810.488.2−77.818th
Wording traps1620.284.5−64.318th
Typos kept on purpose3736.383.7−47.317th

Extraction challenges

GLM 5.3 Flash by extraction challenges
Kind of resume# of resumesF1MedianGapRank
Many entries8343.181.9−38.818th
Long bullet lists10142.881.4−38.718th
Split across pages8551.282.2−31.018th
Big skills list6064.083.6−19.717th
Custom sections10548.280.8−32.618th
Combined headings3765.782.7−17.016th
Promotions8859.283.5−24.317th
Partial dates13852.283.4−31.318th
Overlapping jobs6950.481.7−31.318th
Accented text1351.185.5−34.418th

Noise

GLM 5.3 Flash by noise
Kind of resume# of resumesF1MedianGapRank
Repeated header6766.185.2−19.016th
References5552.185.2−33.118th
Busy page1848.280.1−31.918th

Source

GLM 5.3 Flash by source
Kind of resume# of resumesF1MedianGapRank
Everyday resumes21266.385.1−18.817th
Expert-built (Canva / LaTeX)3242.983.6−40.818th
Resume-builder templates1048.881.2−32.318th
Scanned & rotated1954.882.9−28.116th
Distractor layouts1817.885.3−67.518th

Layout

GLM 5.3 Flash by layout
Kind of resume# of resumesF1MedianGapRank
Single column19557.084.0−26.918th
Two-column2362.686.8−24.318th
Two-column sidebar5461.280.0−18.816th
Three-column2062.084.8−22.818th
Table grid883.094.1−11.116th

Length

GLM 5.3 Flash by length
Kind of resume# of resumesF1MedianGapRank
Short14879.987.1−7.214th
Medium9024.284.7−60.518th
Long6261.274.6−13.416th

By section of the resume

F1 by resume section, against the median system.

GLM 5.3 Flash by resume section
SectionResumesF1MedianGap
Contact details29873.393.5−20.3
Experience29855.486.3−30.9
Education29664.289.4−25.2
Summary28850.781.7−31.0
Certifications24457.381.1−23.8
Awards10354.375.4−21.1
Skills28566.092.1−26.1
Languages17865.884.3−18.5
Custom sections21232.049.9−17.9
Interests7444.391.7−47.4
Volunteering7943.084.5−41.6
Profiles3841.266.7−25.5
Projects8160.583.2−22.7
Publications5065.078.6−13.6

On the published resumes

Best and worst of the 20 public resumes.

GLM 5.3 Flash on published resumes
ResumeWhat makes it hardF1
Creative DirectorAccented text in a styled CV91.4
Lead Respiratory Therapist (Adult ICU)Long resume, 34 entries89.0
Deputy Head of MissionRotated scan with hidden identity text87.0
Head of Events & Public Space OperationsTable grid with emoji bullets10.8
PartnerConsulting CV, six pages10.3
Quality Systems LeadPhoto of a printout9.2

How it was run

Z.ai API, GLM-5.3-Flash, page images in, JSON schema out. Same PDF, schema and instructions as every other system, one run per resume (4 November 2026).

GLM 5.3 Flash on Zhipu AI’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.

GLM 5.3 Flash: #18 on ResumeBench resume extraction benchmark
<a href="https://www.resumebench.ai/systems/glm-5-3-flash"><img src="https://www.resumebench.ai/badge/glm-5-3-flash.svg" alt="GLM 5.3 Flash: #18 on ResumeBench resume extraction benchmark" height="28"></a>

Choosing a resume extraction model

All 18 models · Leaderboard

Frequently asked questions (FAQs)

How accurate is GLM 5.3 Flash at resume extraction?

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

How much does GLM 5.3 Flash cost to parse resumes?

$0.80 per 1,000 pages, about $0.002 per resume.

How fast is GLM 5.3 Flash?

46 seconds per resume (median).

Which resumes is GLM 5.3 Flash best at?

Handwritten, short and table grid. Weakest on content as images, distractor layouts and wording traps.

What is a cheaper alternative to GLM 5.3 Flash?

Gemma 4 26B: 62.8 F1 for $0.56 per 1,000 pages.