Gemma 4 26B vs GPT-5.4 Nano: which is better at resume extraction?
Gemma 4 26B leads GPT-5.4 Nano by 1.0 points (62.8 vs 61.9 F1). Cost per 1,000 pages: $0.56 vs $2.76. Median time per resume: 29 s vs 20 s.
Gemma 4 26B wins on 154 of 300 resumes, GPT-5.4 Nano on 145, 1 tied.
62.82
Gemma 4 26B
A · Vision LLM
61.86
GPT-5.4 Nano
B · Vision LLM
Gemma 4 26B is ahead by 0.96 points on these 300 resumes.
- Many entries 83A 58.0 · B 59.7−1.7
- Long bullet lists 101A 57.4 · B 61.0−3.6
- Split across pages 85A 60.0 · B 61.7−1.7
- Big skills list 60A 61.9 · B 64.9−3.0
- Custom sections 105A 58.9 · B 59.5−0.6
- Combined headings 37A 60.9 · B 64.6−3.7
- Promotions 88A 59.5 · B 58.7+0.8
- Partial dates 138A 62.9 · B 62.6+0.3
- Overlapping jobs 69A 57.5 · B 58.6−1.2
- Accented text 13A 59.3 · B 55.9+3.4
Side by side
| Gemma 4 26B | GPT-5.4 Nano | |
|---|---|---|
| Type | Vision LLM | Vision LLM |
| Maker | OpenAI | |
| Rank | 16 of 18 | 17 of 18 |
| F1 | 62.8 | 61.9 |
| Precision | 69.1 | 70.1 |
| Recall | 66.2 | 66.3 |
| Cost per 1,000 pages | $0.56 | $2.76 |
| Median time per resume | 29 s | 20 s |
| Resumes won | 154 | 145 |
Where each one wins
Where the two differ most.
| Kind of resume | # of resumes | Gemma 4 26B | GPT-5.4 Nano | Gap |
|---|---|---|---|---|
| Scanned & rotated (source) | 19 | 40.7 | 6.3 | +34.4 |
| Rotated & warped pages (distractors) | 21 | 43.4 | 11.7 | +31.7 |
| Hidden text & prompt injections (distractors) | 14 | 48.6 | 22.0 | +26.6 |
| Image only (reading the page) | 28 | 45.1 | 21.0 | +24.2 |
| Typos kept on purpose (distractors) | 37 | 51.3 | 33.5 | +17.7 |
| Expert-built (Canva / LaTeX) (source) | 32 | 70.2 | 60.3 | +9.9 |
| Resume-builder templates (source) | 10 | 58.6 | 67.6 | −8.9 |
| Two-column sidebar (layout) | 54 | 55.1 | 47.4 | +7.7 |
| Photo (reading the page) | 45 | 66.9 | 59.4 | +7.5 |
| Sidebar or grid (reading the page) | 82 | 60.4 | 54.9 | +5.6 |
| Table grid (layout) | 8 | 79.8 | 74.4 | +5.4 |
| References (noise) | 55 | 62.5 | 67.1 | −4.6 |
| Section | Resumes | Gemma 4 26B | GPT-5.4 Nano | Gap |
|---|---|---|---|---|
| Contact details | 300 | 84.6 | 81.2 | +3.4 |
| Experience | 300 | 56.0 | 63.7 | −7.7 |
| Education | 297 | 68.0 | 62.2 | +5.8 |
| Summary | 290 | 35.5 | 71.7 | −36.2 |
| Certifications | 244 | 63.0 | 53.6 | +9.4 |
| Awards | 110 | 55.7 | 44.9 | +10.8 |
| Skills | 290 | 81.9 | 74.8 | +7.1 |
| Languages | 171 | 80.9 | 69.0 | +11.9 |
| Custom sections | 223 | 32.3 | 26.3 | +5.9 |
| Interests | 67 | 71.8 | 70.0 | +1.8 |
| Volunteering | 78 | 59.4 | 59.9 | −0.5 |
| Profiles | 37 | 57.2 | 23.7 | +33.5 |
| Projects | 83 | 55.0 | 48.8 | +6.2 |
| Publications | 53 | 54.5 | 30.7 | +23.8 |
Published resumes they disagree on most
Open one to see both outputs field by field.
| Resume | Gemma 4 26B | GPT-5.4 Nano | Gap |
|---|---|---|---|
| Partner: Consulting CV, six pages | 70.3 | 27.8 | +42.5 |
| Deputy Head of Mission: Rotated scan with hidden identity text | 34.1 | 1.7 | +32.4 |
| Head of Digital Marketing: Canva design with a photo | 69.4 | 39.3 | +30.1 |
| Creative Director: Accented text in a styled CV | 39.3 | 63.6 | −24.3 |
Which to pick
Gemma 4 26B is more accurate. GPT-5.4 Nano is the faster of the two.
Gemma 4 26B does better on scanned & rotated, rotated & warped pages and hidden text & prompt injections. GPT-5.4 Nano does better on resume-builder templates, references and long.
Gemma 4 26B against every other model
- Gemma 4 26B vs LlamaParse Extract Agentic Plus
- Gemma 4 26B vs LlamaParse Extract Agentic
- Gemma 4 26B vs LlamaParse Extract Cost-Effective
- Gemma 4 26B vs Reducto Deep Extract
- Gemma 4 26B vs Reducto Extract
- Gemma 4 26B vs Extend Max Context
- Gemma 4 26B vs Extend Extract
- Gemma 4 26B vs GPT-5.6
- Gemma 4 26B vs GPT-5.5
- Gemma 4 26B vs Claude Opus 4.8
- Gemma 4 26B vs Gemini 3.5 Flash
- Gemma 4 26B vs Qwen 3.8 Flash
- Gemma 4 26B vs Qwen 3.8 27B
- Gemma 4 26B vs Qwen 3.5 35B
- Gemma 4 26B vs Qwen 3.5 9B
- Gemma 4 26B 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
- GPT-5.4 Nano vs Qwen 3.5 35B
- GPT-5.4 Nano vs Qwen 3.5 9B
- GPT-5.4 Nano vs GLM 5.3 Flash
Popular comparisons
- LlamaParse Extract Agentic Plus vs Reducto Deep Extract
- Extend Max Context vs Reducto Deep Extract
- Extend Max Context vs LlamaParse Extract Agentic Plus
- Claude Opus 4.8 vs GPT-5.6
- Gemini 3.5 Flash vs GPT-5.6
- Claude Opus 4.8 vs Gemini 3.5 Flash
- LlamaParse Extract Agentic vs LlamaParse Extract Agentic Plus
- Reducto Deep Extract vs Reducto Extract
- Gemma 4 26B vs Qwen 3.8 27B
- Document APIs vs vision LLMs