Extend · Document API
Extend Max Context on ResumeBench: 84.7 F1, rank 7 of 18
Extend Max Context ranks 7th of 18 on ResumeBench: 84.7 F1, $50 per 1,000 pages, 43 seconds per resume. It is 2.4 points behind LlamaParse Extract Agentic Plus. Reducto Deep Extract scores 85.6 for $40.
4th of 7 document APIs. Strongest on scanned & rotated (+8.5 points), rotated & warped pages (+7.8 points) and typos kept on purpose (+5.6 points). Weakest on handwritten (−4.9 points), combined headings (−2.3 points) and two-column (−1.5 points).
- Rank
- 7/18
- F1
- 84.7
- Precision
- 86.6
- Recall
- 86.4
- Perfect resumes
- 35/300
- Per 1,000 pages
- $50
- Median per resume
- 43 s
Averaged over 300 resumes. Dates (scored separately): 97.8. 90% finish within 82 s. Cost from list price.
Where Extend Max Context 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 | 46.5 | 51.5 | −4.9 | 17th |
| Image only | 28 | 76.5 | 71.4 | +5.1 | 4th |
| Sidebar or grid | 82 | 84.4 | 81.9 | +2.4 | 5th |
| Photo | 45 | 85.6 | 84.0 | +1.5 | 5th |
Distractors
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Rotated & warped pages | 21 | 90.8 | 82.9 | +7.8 | 3rd |
| Hidden text & prompt injections | 14 | 89.6 | 84.5 | +5.1 | 6th |
| Content as images | 8 | 89.4 | 88.2 | +1.2 | 7th |
| Wording traps | 16 | 85.6 | 84.5 | +1.1 | 6th |
| Typos kept on purpose | 37 | 89.3 | 83.7 | +5.6 | 3rd |
Extraction challenges
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Many entries | 83 | 85.6 | 81.9 | +3.7 | 1st |
| Long bullet lists | 101 | 84.8 | 81.4 | +3.3 | 3rd |
| Split across pages | 85 | 83.1 | 82.2 | +0.9 | 8th |
| Big skills list | 60 | 85.8 | 83.6 | +2.2 | 3rd |
| Custom sections | 105 | 82.0 | 80.8 | +1.2 | 7th |
| Combined headings | 37 | 80.4 | 82.7 | −2.3 | 11th |
| Promotions | 88 | 85.8 | 83.5 | +2.3 | 4th |
| Partial dates | 138 | 84.8 | 83.4 | +1.4 | 6th |
| Overlapping jobs | 69 | 85.2 | 81.7 | +3.5 | 3rd |
| Accented text | 13 | 88.8 | 85.5 | +3.3 | 3rd |
Noise
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Repeated header | 67 | 85.5 | 85.2 | +0.3 | 8th |
| References | 55 | 90.3 | 85.2 | +5.1 | 2nd |
| Busy page | 18 | 83.2 | 80.1 | +3.1 | 4th |
Source
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Everyday resumes | 212 | 85.7 | 85.1 | +0.6 | 8th |
| Expert-built (Canva / LaTeX) | 32 | 84.8 | 83.6 | +1.2 | 4th |
| Resume-builder templates | 10 | 80.6 | 81.2 | −0.6 | 10th |
| Scanned & rotated | 19 | 91.4 | 82.9 | +8.5 | 3rd |
| Distractor layouts | 18 | 87.0 | 85.3 | +1.7 | 5th |
Layout
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Single column | 195 | 84.8 | 84.0 | +0.8 | 7th |
| Two-column | 23 | 85.3 | 86.8 | −1.5 | 11th |
| Two-column sidebar | 54 | 83.0 | 80.0 | +3.0 | 5th |
| Three-column | 20 | 84.8 | 84.8 | −0.1 | 10th |
| Table grid | 8 | 92.6 | 94.1 | −1.5 | 13th |
Length
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Short | 148 | 87.9 | 87.1 | +0.7 | 6th |
| Medium | 90 | 85.1 | 84.7 | +0.4 | 7th |
| Long | 62 | 76.7 | 74.6 | +2.1 | 8th |
By section of the resume
F1 by resume section, against the median system.
| Section | Resumes | F1 | Median | Gap |
|---|---|---|---|---|
| Contact details | 300 | 95.5 | 93.5 | +2.0 |
| Experience | 300 | 87.9 | 86.3 | +1.6 |
| Education | 297 | 91.2 | 89.4 | +1.7 |
| Summary | 290 | 87.6 | 81.7 | +5.9 |
| Certifications | 251 | 81.8 | 81.1 | +0.7 |
| Awards | 92 | 78.6 | 75.4 | +3.2 |
| Skills | 285 | 88.7 | 92.1 | −3.4 |
| Languages | 167 | 84.6 | 84.3 | +0.3 |
| Custom sections | 212 | 50.2 | 49.9 | +0.3 |
| Interests | 64 | 91.3 | 91.7 | −0.4 |
| Volunteering | 76 | 87.3 | 84.5 | +2.8 |
| Profiles | 34 | 69.6 | 66.7 | +2.9 |
| Projects | 80 | 84.5 | 83.2 | +1.3 |
| Publications | 51 | 91.2 | 78.6 | +12.6 |
On the published resumes
Best and worst of the 20 public resumes.
| Resume | What makes it hard | F1 |
|---|---|---|
| Senior Controls Engineer | Five custom sections | 96.2 |
| Senior Data Scientist (Lead) | Fifty skills, seven pages | 95.9 |
| Senior Manufacturing Engineer | Third-person prose, no bullets | 92.1 |
| Senior Game Designer | Game designer's Canva resume | 66.4 |
| senior Consultant - package Implementation | Handwritten resume | 28.6 |
| Senior ocularist and practice owner | Handwritten resume, six pages | 26.9 |
How it was run
Extend extraction with the large_array_max_context array strategy. Same PDF, schema and instructions as every other system, one run per resume (4 November 2026).
Max Context makes several passes through the document for long lists (the large_array_max_context strategy), at about twice the credits.
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 testClosest alternatives
Nearest in F1.
| Model | F1 | Per 1,000 pages | Median time | Head to head |
|---|---|---|---|---|
| Claude Opus 4.8 | 85.5 | $90 | 31 s | Extend Max Context vs Claude Opus 4.8 |
| Gemini 3.5 Flash | 85.5 | $35 | 43 s | Extend Max Context vs Gemini 3.5 Flash |
| GPT-5.6 | 84.7 | $74 | 38 s | Extend Max Context vs GPT-5.6 |
| Qwen 3.8 Flash | 84.1 | $2.98 | 122 s | Extend Max Context vs Qwen 3.8 Flash |
Extend Max Context against every other model
- Extend Max Context vs LlamaParse Extract Agentic Plus
- Extend Max Context vs LlamaParse Extract Agentic
- Extend Max Context vs LlamaParse Extract Cost-Effective
- Extend Max Context vs Reducto Deep Extract
- Extend Max Context vs Reducto Extract
- Extend Max Context vs Extend Extract
- Extend Max Context vs GPT-5.5
- Extend Max Context vs GPT-5.4 Nano
- Extend Max Context vs Gemma 4 26B
- Extend Max Context vs Qwen 3.8 27B
- Extend Max Context vs Qwen 3.5 35B
- Extend Max Context vs Qwen 3.5 9B
- Extend Max Context vs GLM 5.3 Flash
Embed this result
Shows its current rank. Updates with each run.