llm-benchmarks
githubdrose.io

PaddlePaddle/PaddleOCR-VL-0.9B Benchmarks


Measured on deepinfra

2 runs
deepinfra site
Avg Tokens / Second
5.90
Avg Time to First Token (ms)
0.00
Runs Analysed
2
Last Updated
Aug 5, 2026, 04:48 PM

Throughput distribution

  • PaddlePaddle/PaddleOCR-VL-0.9B deepinfra6 tok/s
025

Visible-token throughput, same basis as the table · curves normalised per model

Throughput over time

PaddlePaddle/PaddleOCR-VL-0.9Bdeepinfra0tok/s

Generated throughput, including reasoning tokens — so a thinking model's line does not collapse; the table ranks on visible tokens · one cell per model, showing its best-covered provider · shared vertical scale, 0 to 1 tok/s (99th percentile) · dashed rule is the model's own mean · 1 samples above the ceiling drawn at it

Every provider serving this model

1 measured
ProviderModelAvg Toks/SecMinMaxAvg TTF (ms)
deepinfraPaddlePaddle/PaddleOCR-VL-0.9B5.904.267.530.00

Frequently Asked Questions

How fast is PaddlePaddle/PaddleOCR-VL-0.9B?
The latest rolling average throughput is 5.90 tokens per second with an average time to first token of 0.00 ms across 2 recent runs.
How often are these benchmarks updated?
Benchmarks refresh automatically whenever the monitoring cron runs. The most recent run completed on Aug 5, 2026, 04:48 PM.