Which resume extraction model handles scanned and image-only resumes best?
28 of the 300 resumes are scanned and image-only resumes. The typical system scores 10.5 points lower on them. Best: LlamaParse Extract Agentic Plus, 84.2 F1.
PDFs with no text layer: scans, photos of printouts and exported images, where the words have to be read from pixels. Best document API: LlamaParse Extract Agentic Plus (84.2). Best vision LLM: Claude Opus 4.8 (77.0). LlamaParse Extract Cost-Effective ranks 4 places higher here than overall.
Every system on scanned and image-only resumes
F1 on the 28 scanned and image-only resumes, against each system’s F1 on all 300 resumes. Cost and time are for all resumes.
| Model | F1 here | F1 overall | Gap | Per 1,000 pages | Median time |
|---|---|---|---|---|---|
| LlamaParse Extract Agentic Plus | 84.2 | 87.2 | −3.0 | $54 | 70 s |
| LlamaParse Extract Agentic | 81.5 | 85.5 | −4.1 | $23 | 55 s |
| Claude Opus 4.8 | 77.0 | 85.5 | −8.5 | $90 | 31 s |
| Extend Max Context | 76.5 | 84.7 | −8.3 | $50 | 43 s |
| Reducto Deep Extract | 76.4 | 85.6 | −9.1 | $40 | 65 s |
| LlamaParse Extract Cost-Effective | 76.3 | 83.5 | −7.2 | $8.18 | 29 s |
| Gemini 3.5 Flash | 75.1 | 85.5 | −10.3 | $35 | 43 s |
| Qwen 3.8 Flash | 74.7 | 84.1 | −9.4 | $2.98 | 122 s |
| Reducto Extract | 72.0 | 81.7 | −9.7 | $20 | 16 s |
| Extend Extract | 70.7 | 81.0 | −10.2 | $12 | 31 s |
| Qwen 3.8 27B | 70.5 | 83.8 | −13.2 | $9.59 | 36 s |
| Qwen 3.5 35B | 69.7 | 77.9 | −8.2 | $3.91 | 78 s |
| GPT-5.6 | 65.4 | 84.7 | −19.3 | $74 | 38 s |
| Qwen 3.5 9B | 61.2 | 76.0 | −14.8 | $1.31 | 138 s |
| GPT-5.5 | 57.0 | 83.5 | −26.5 | $87 | 48 s |
| GLM 5.3 Flash | 53.3 | 59.2 | −6.0 | $0.80 | 46 s |
| Gemma 4 26B | 45.1 | 62.8 | −17.7 | $0.56 | 29 s |
| GPT-5.4 Nano | 21.0 | 61.9 | −40.9 | $2.76 | 20 s |
What to do about it
If you get a lot of scans, rank parsers on this table, not the overall one.
Published examples
4 public resumes of this kind.