LlamaIndex · Document API
LlamaParse Extract Agentic Plus on ResumeBench: 87.2 F1, rank 1 of 18
LlamaParse Extract Agentic Plus ranks 1st of 18 on ResumeBench: 87.2 F1, $54 per 1,000 pages, 70 seconds per resume. LlamaParse Extract Agentic scores 85.5 for $23, 57% less.
1st of 7 document APIs. Strongest on handwritten (+16.4 points), image only (+12.8 points) and scanned & rotated (+9.8 points).
- Rank
- 1/18
- F1
- 87.2
- Precision
- 88.4
- Recall
- 88.3
- Perfect resumes
- 46/300
- Per 1,000 pages
- $54
- Median per resume
- 70 s
Averaged over 300 resumes. Dates (scored separately): 98.3. 90% finish within 164 s.
Where LlamaParse Extract Agentic Plus wins and loses
F1 by kind of resume, against the median system.
Reading the page
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Handwritten | 9 | 67.8 | 51.5 | +16.4 | 1st |
| Image only | 28 | 84.2 | 71.4 | +12.8 | 1st |
| Sidebar or grid | 82 | 86.8 | 81.9 | +4.8 | 1st |
| Photo | 45 | 88.1 | 84.0 | +4.1 | 1st |
Distractors
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Rotated & warped pages | 21 | 92.5 | 82.9 | +9.5 | 1st |
| Hidden text & prompt injections | 14 | 93.2 | 84.5 | +8.7 | 1st |
| Content as images | 8 | 93.9 | 88.2 | +5.7 | 1st |
| Wording traps | 16 | 89.0 | 84.5 | +4.5 | 2nd |
| Typos kept on purpose | 37 | 91.4 | 83.7 | +7.7 | 1st |
Extraction challenges
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Many entries | 83 | 85.0 | 81.9 | +3.2 | 2nd |
| Long bullet lists | 101 | 84.9 | 81.4 | +3.5 | 1st |
| Split across pages | 85 | 85.3 | 82.2 | +3.1 | 1st |
| Big skills list | 60 | 86.0 | 83.6 | +2.4 | 2nd |
| Custom sections | 105 | 85.1 | 80.8 | +4.3 | 1st |
| Combined headings | 37 | 84.6 | 82.7 | +1.9 | 2nd |
| Promotions | 88 | 87.4 | 83.5 | +3.9 | 1st |
| Partial dates | 138 | 87.0 | 83.4 | +3.5 | 1st |
| Overlapping jobs | 69 | 84.1 | 81.7 | +2.3 | 4th |
| Accented text | 13 | 92.0 | 85.5 | +6.5 | 1st |
Noise
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Repeated header | 67 | 89.3 | 85.2 | +4.1 | 1st |
| References | 55 | 87.6 | 85.2 | +2.4 | 7th |
| Busy page | 18 | 82.8 | 80.1 | +2.7 | 5th |
Source
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Everyday resumes | 212 | 87.4 | 85.1 | +2.3 | 1st |
| Expert-built (Canva / LaTeX) | 32 | 86.4 | 83.6 | +2.7 | 1st |
| Resume-builder templates | 10 | 85.6 | 81.2 | +4.4 | 3rd |
| Scanned & rotated | 19 | 92.8 | 82.9 | +9.8 | 1st |
| Distractor layouts | 18 | 89.9 | 85.3 | +4.6 | 2nd |
Layout
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Single column | 195 | 87.0 | 84.0 | +3.0 | 1st |
| Two-column | 23 | 90.3 | 86.8 | +3.4 | 1st |
| Two-column sidebar | 54 | 84.8 | 80.0 | +4.8 | 1st |
| Three-column | 20 | 87.6 | 84.8 | +2.8 | 1st |
| Table grid | 8 | 98.1 | 94.1 | +4.0 | 1st |
Length
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Short | 148 | 90.1 | 87.1 | +3.0 | 1st |
| Medium | 90 | 87.4 | 84.7 | +2.7 | 1st |
| Long | 62 | 79.7 | 74.6 | +5.1 | 1st |
By section of the resume
F1 by resume section, against the median system.
| Section | Resumes | F1 | Median | Gap |
|---|---|---|---|---|
| Contact details | 300 | 94.3 | 93.5 | +0.8 |
| Experience | 300 | 91.5 | 86.3 | +5.2 |
| Education | 297 | 93.3 | 89.4 | +3.9 |
| Summary | 290 | 84.8 | 81.7 | +3.1 |
| Certifications | 240 | 85.8 | 81.1 | +4.7 |
| Awards | 93 | 84.1 | 75.4 | +8.7 |
| Skills | 283 | 97.6 | 92.1 | +5.5 |
| Languages | 167 | 84.6 | 84.3 | +0.3 |
| Custom sections | 211 | 52.2 | 49.9 | +2.3 |
| Interests | 64 | 93.0 | 91.7 | +1.3 |
| Volunteering | 77 | 89.6 | 84.5 | +5.0 |
| Profiles | 35 | 73.3 | 66.7 | +6.6 |
| Projects | 78 | 87.0 | 83.2 | +3.8 |
| Publications | 52 | 85.5 | 78.6 | +7.0 |
On the published resumes
Best and worst of the 20 public resumes.
| Resume | What makes it hard | F1 |
|---|---|---|
| Head of Events & Public Space Operations | Table grid with emoji bullets | 98.4 |
| Senior Data Scientist (Lead) | Fifty skills, seven pages | 97.7 |
| Head of Developer Experience (Internal title: Jedi Master of Analytics) | Unusual job titles | 96.3 |
| Senior Full-Stack Engineer | LaTeX CV with a running header | 77.0 |
| Senior ocularist and practice owner | Handwritten resume, six pages | 74.0 |
| Senior Game Designer | Game designer's Canva resume | 64.8 |
How it was run
LlamaParse Extract API, extraction mode agentic_plus. Same PDF, schema and instructions as every other system, one run per resume (4 November 2026).
LlamaParse Extract (formerly LlamaExtract) is the structured-extraction API of LlamaIndex’s LlamaParse platform.
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 llamaextract_agentic_plus --split test
$ resume-bench grade llamaextract_agentic_plus --split testClosest alternatives
Nearest in F1.
| Model | F1 | Per 1,000 pages | Median time | Head to head |
|---|---|---|---|---|
| Reducto Deep Extract | 85.6 | $40 | 65 s | LlamaParse Extract Agentic Plus vs Reducto Deep Extract |
| LlamaParse Extract Agentic | 85.5 | $23 | 55 s | LlamaParse Extract Agentic Plus vs LlamaParse Extract Agentic |
| Claude Opus 4.8 | 85.5 | $90 | 31 s | LlamaParse Extract Agentic Plus vs Claude Opus 4.8 |
| Gemini 3.5 Flash | 85.5 | $35 | 43 s | LlamaParse Extract Agentic Plus vs Gemini 3.5 Flash |
LlamaParse Extract Agentic Plus against every other model
- LlamaParse Extract Agentic Plus vs LlamaParse Extract Cost-Effective
- LlamaParse Extract Agentic Plus vs Reducto Extract
- LlamaParse Extract Agentic Plus vs Extend Max Context
- LlamaParse Extract Agentic Plus vs Extend Extract
- LlamaParse Extract Agentic Plus vs GPT-5.6
- LlamaParse Extract Agentic Plus vs GPT-5.5
- LlamaParse Extract Agentic Plus vs GPT-5.4 Nano
- LlamaParse Extract Agentic Plus vs Gemma 4 26B
- LlamaParse Extract Agentic Plus vs Qwen 3.8 Flash
- LlamaParse Extract Agentic Plus vs Qwen 3.8 27B
- LlamaParse Extract Agentic Plus vs Qwen 3.5 35B
- LlamaParse Extract Agentic Plus vs Qwen 3.5 9B
- LlamaParse Extract Agentic Plus vs GLM 5.3 Flash
Embed this result
Shows its current rank. Updates with each run.