multimodal benchmark
SVG-Bench
SVG-Bench is an internal benchmark that comprehensively evaluates SVG generation performance. It accepts text and image inputs across build-from-scratch and edit-based tasks, using a VLM to verify rendering accuracy of the generated outputs.
Updated Aug 11, 2026
Models1
Model coverage1
MetricScore
EvidenceB
SVG-Bench Ranking
Higher score ranks better on this benchmark.
| 1 | MI | 63.7% | 100.0% | 1 | C |
SVG-Bench Highlights
The leading models and scores on this benchmark.
What is SVG-Bench?
What SVG-Bench measures and how its scores work.
SVG-Bench is an internal benchmark that comprehensively evaluates SVG generation performance. It accepts text and image inputs across build-from-scratch and edit-based tasks, using a VLM to verify rendering accuracy of the generated outputs.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- SVG-Bench
- Modality
- multimodal
- Primary category
- multimodal
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- svg-bench|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about SVG-Bench.
Which model scores highest on SVG-Bench?
MiniMax M3 is currently ranked first with 63.7%.
What does SVG-Bench measure?
SVG-Bench is an internal benchmark that comprehensively evaluates SVG generation performance. It accepts text and image inputs across build-from-scratch and edit-based tasks, using a VLM to verify rendering accuracy of the generated outputs.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
1 model results are currently shown.
Does this benchmark affect the overall score?
No. This benchmark is shown for reference but does not contribute to the overall score.