Model comparison
Compare models or reasoning efforts by score, evaluation time and estimated cost.
Available modelsSelected 2 / 3
Measured results
Differences are relative to the first column
deepseek-v4-proFrontend −24 pts · Reasoning +15 pts
| Metric | Qwen 3.8 Flash · default | deepseek-v4-pro · max |
|---|---|---|
| Overall | 65.15 | 65.15Tied |
| Backend engineering | 68 | 67−1 pts |
| Frontend interaction | 87 | 63−24 pts |
| Knowledge & reasoning | 50 | 65+15 pts |
| Visual aesthetics | 73 | 70−3 pts |
| Maintainability | 57 | 59+2 pts |
Cost & time
Complete evaluation| Metric | Qwen 3.8 Flash · default | deepseek-v4-pro · max |
|---|---|---|
| Reference cost | $0.42 | $1.15+$0.74 |
| Evaluation time | 134m 31s | 93m 15s−30.7% |
Dimension breakdown
Expand to inspect each subaxis.
Backend engineeringSubscores
| Scoring dimension | Qwen 3.8 Flashdefault | deepseek-v4-promax |
|---|---|---|
| Contracts and boundaries | 8 / 20 | 9 / 20 |
| Retries and failures | 15 / 20 | 17 / 20 |
| CI and adversarial review | 13 / 20 | 12 / 20 |
| Transactions and consistency | 15 / 20 | 13 / 20 |
| Caching and regression | 17 / 20 | 16 / 20 |
Frontend interactionSubscores
| Scoring dimension | Qwen 3.8 Flashdefault | deepseek-v4-promax |
|---|---|---|
| Layout and adaptation | 20 / 20 | 17 / 20 |
| Interaction behavior | 19 / 20 | 16 / 20 |
| State continuity | 18 / 20 | 8 / 20 |
| Accessibility and feedback | 15 / 15 | 7 / 15 |
| State validation and regression | 15 / 25 | 15 / 25 |
Knowledge & reasoningSubscores
| Scoring dimension | Qwen 3.8 Flashdefault | deepseek-v4-promax |
|---|---|---|
| Mathematical reasoning | 15 / 15 | 15 / 15 |
| Computer science & AI | 10 / 15 | 10 / 15 |
| Science and engineering | 5 / 10 | 5 / 10 |
| Life sciences and medicine | 10 / 20 | 15 / 20 |
| Humanities and general knowledge | 10 / 40 | 20 / 40 |
Visual aestheticsSubscores
| Scoring dimension | Qwen 3.8 Flashdefault | deepseek-v4-promax |
|---|---|---|
| Composition | 18 / 20 | 19 / 20 |
| Spacing and alignment | 18 / 20 | 16 / 20 |
| Typography and expression | 17 / 20 | 18 / 20 |
| Color and surfaces | 20 / 20 | 19 / 20 |
| Control styling | 17 / 20 | 17 / 20 |
Subscores retain their original scale; the axis score uses a consistent conversion.
MaintainabilitySubscores
| Scoring dimension | Qwen 3.8 Flashdefault | deepseek-v4-promax |
|---|---|---|
| Semantic clarity | 15 / 20 | 15 / 20 |
| Control-flow organization | 20 / 20 | 20 / 20 |
| Locality of changes | 15 / 20 | 15 / 20 |
| State and lifecycle | 16 / 20 | 17 / 20 |
| Implementation economy | 17 / 20 | 17 / 20 |
Subscores retain their original scale; the axis score uses a consistent conversion.