OpenAI · Vision LLM
GPT-5.4 Nano on ResumeBench: 61.9 F1, rank 17 of 18
GPT-5.4 Nano ranks 17th of 18 on ResumeBench: 61.9 F1, $2.76 per 1,000 pages, 20 seconds per resume. It is 25.3 points behind LlamaParse Extract Agentic Plus. Qwen 3.5 9B scores 76.0 for $1.31.
10th of 11 vision LLMs. Strongest on handwritten (−0.1 points), resume-builder templates (−13.6 points) and long (−15.6 points). Weakest on scanned & rotated (−76.6 points), rotated & warped pages (−71.3 points) and hidden text & prompt injections (−62.6 points).
- Rank
- 17/18
- F1
- 61.9
- Precision
- 70.1
- Recall
- 66.3
- Perfect resumes
- 1/300
- Per 1,000 pages
- $2.76
- Median per resume
- 20 s
Averaged over 300 resumes. Dates (scored separately): 86.7. 90% finish within 34 s.
Where GPT-5.4 Nano 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 | 51.3 | 51.5 | −0.1 | 10th |
| Image only | 28 | 21.0 | 71.4 | −50.4 | 18th |
| Sidebar or grid | 82 | 54.9 | 81.9 | −27.1 | 18th |
| Photo | 45 | 59.4 | 84.0 | −24.7 | 18th |
Distractors
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Rotated & warped pages | 21 | 11.7 | 82.9 | −71.3 | 18th |
| Hidden text & prompt injections | 14 | 22.0 | 84.5 | −62.6 | 18th |
| Content as images | 8 | 61.3 | 88.2 | −26.9 | 17th |
| Wording traps | 16 | 62.8 | 84.5 | −21.7 | 16th |
| Typos kept on purpose | 37 | 33.5 | 83.7 | −50.1 | 18th |
Extraction challenges
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Many entries | 83 | 59.7 | 81.9 | −22.2 | 16th |
| Long bullet lists | 101 | 61.0 | 81.4 | −20.4 | 16th |
| Split across pages | 85 | 61.7 | 82.2 | −20.5 | 16th |
| Big skills list | 60 | 64.9 | 83.6 | −18.7 | 16th |
| Custom sections | 105 | 59.5 | 80.8 | −21.3 | 16th |
| Combined headings | 37 | 64.6 | 82.7 | −18.1 | 17th |
| Promotions | 88 | 58.7 | 83.5 | −24.8 | 18th |
| Partial dates | 138 | 62.6 | 83.4 | −20.9 | 17th |
| Overlapping jobs | 69 | 58.6 | 81.7 | −23.1 | 16th |
| Accented text | 13 | 55.9 | 85.5 | −29.6 | 17th |
Noise
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Repeated header | 67 | 64.2 | 85.2 | −21.0 | 17th |
| References | 55 | 67.1 | 85.2 | −18.1 | 16th |
| Busy page | 18 | 61.6 | 80.1 | −18.5 | 16th |
Source
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Everyday resumes | 212 | 67.2 | 85.1 | −17.9 | 16th |
| Expert-built (Canva / LaTeX) | 32 | 60.3 | 83.6 | −23.3 | 17th |
| Resume-builder templates | 10 | 67.6 | 81.2 | −13.6 | 16th |
| Scanned & rotated | 19 | 6.3 | 82.9 | −76.6 | 18th |
| Distractor layouts | 18 | 62.2 | 85.3 | −23.1 | 17th |
Layout
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Single column | 195 | 63.8 | 84.0 | −20.2 | 16th |
| Two-column | 23 | 70.2 | 86.8 | −16.6 | 16th |
| Two-column sidebar | 54 | 47.4 | 80.0 | −32.6 | 18th |
| Three-column | 20 | 67.2 | 84.8 | −17.6 | 16th |
| Table grid | 8 | 74.4 | 94.1 | −19.7 | 18th |
Length
| Kind of resume | # of resumes | F1 | Median | Gap | Rank |
|---|---|---|---|---|---|
| Short | 148 | 64.1 | 87.1 | −23.1 | 18th |
| Medium | 90 | 60.2 | 84.7 | −24.5 | 17th |
| Long | 62 | 59.0 | 74.6 | −15.6 | 17th |
By section of the resume
F1 by resume section, against the median system.
| Section | Resumes | F1 | Median | Gap |
|---|---|---|---|---|
| Contact details | 300 | 81.2 | 93.5 | −12.4 |
| Experience | 300 | 63.7 | 86.3 | −22.6 |
| Education | 297 | 62.2 | 89.4 | −27.2 |
| Summary | 290 | 71.7 | 81.7 | −10.0 |
| Certifications | 244 | 53.6 | 81.1 | −27.5 |
| Awards | 102 | 44.9 | 75.4 | −30.5 |
| Skills | 291 | 74.8 | 92.1 | −17.3 |
| Languages | 170 | 69.0 | 84.3 | −15.3 |
| Custom sections | 231 | 26.3 | 49.9 | −23.6 |
| Interests | 66 | 70.0 | 91.7 | −21.7 |
| Volunteering | 80 | 59.9 | 84.5 | −24.7 |
| Profiles | 66 | 23.7 | 66.7 | −43.0 |
| Projects | 81 | 48.8 | 83.2 | −34.4 |
| Publications | 51 | 30.7 | 78.6 | −47.9 |
On the published resumes
Best and worst of the 20 public resumes.
| Resume | What makes it hard | F1 |
|---|---|---|
| Senior Customer Service Representative | Busy Canva page | 81.0 |
| Lead Solutions Architect | Indian format with CGPA | 77.6 |
| Senior Manufacturing Engineer | Third-person prose, no bullets | 75.2 |
| Supervisory Human Resources Specialist, GS-0201-15 | Resume-builder template | 32.5 |
| Partner | Consulting CV, six pages | 27.8 |
| Deputy Head of Mission | Rotated scan with hidden identity text | 1.7 |
How it was run
OpenAI API, model gpt-5.4-nano, page images in, JSON schema out. Same PDF, schema and instructions as every other system, one run per resume (4 November 2026).
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 |
|---|---|---|---|---|
| Qwen 3.5 35B | 77.9 | $3.91 | 78 s | GPT-5.4 Nano vs Qwen 3.5 35B |
| Qwen 3.5 9B | 76.0 | $1.31 | 138 s | GPT-5.4 Nano vs Qwen 3.5 9B |
| Gemma 4 26B | 62.8 | $0.56 | 29 s | GPT-5.4 Nano vs Gemma 4 26B |
| GLM 5.3 Flash | 59.2 | $0.80 | 46 s | GPT-5.4 Nano vs GLM 5.3 Flash |
GPT-5.4 Nano against every other model
- GPT-5.4 Nano vs LlamaParse Extract Agentic Plus
- GPT-5.4 Nano vs LlamaParse Extract Agentic
- GPT-5.4 Nano vs LlamaParse Extract Cost-Effective
- GPT-5.4 Nano vs Reducto Deep Extract
- GPT-5.4 Nano vs Reducto Extract
- GPT-5.4 Nano vs Extend Max Context
- GPT-5.4 Nano vs Extend Extract
- GPT-5.4 Nano vs GPT-5.6
- GPT-5.4 Nano vs GPT-5.5
- GPT-5.4 Nano vs Claude Opus 4.8
- GPT-5.4 Nano vs Gemini 3.5 Flash
- GPT-5.4 Nano vs Qwen 3.8 Flash
- GPT-5.4 Nano vs Qwen 3.8 27B
Embed this result
Shows its current rank. Updates with each run.