Open Access
Table G.1
Perfect match rates by LLM, query generation method, query difficulty, and with or without function calling (without self-correction).
| Simple | Medium | Hard | ||||
|---|---|---|---|---|---|---|
| Model | w/o Func. call | w/ Func. call | w/o Func. call | w/ Func. call | w/o Func. call | w/ Func. call |
| ID Match (Rows) - Direct | ||||||
| Claude Opus 4.6 | 0.844 ± 0.000 | 0.850 ± 0.014 | 0.250 ± 0.053 | 0.280 ± 0.045 | 0.400 ± 0.000 | 0.460 ± 0.055 |
| Gemini 2.5 Pro | 0.809 ± 0.023 | 0.787 ± 0.014 | 0.360 ± 0.117 | 0.320 ± 0.045 | 0.160 ± 0.070 | 0.200 ± 0.071 |
| Gemini 3 Flash | 0.866 ± 0.015 | 0.769 ± 0.034 | 0.300 ± 0.000 | 0.320 ± 0.042 | 0.300 ± 0.000 | 0.360 ± 0.052 |
| ID Match (Rows) - Step-by-Step | ||||||
| Claude Opus 4.6 | 0.938 ± 0.000 | 0.938 ± 0.000 | 0.380 ± 0.042 | 0.390 ± 0.088 | 0.530 ± 0.048 | 0.460 ± 0.084 |
| Gemini 2.5 Pro | 0.847 ± 0.023 | 0.863 ± 0.017 | 0.490 ± 0.032 | 0.500 ± 0.000 | 0.250 ± 0.053 | 0.240 ± 0.055 |
| Gemini 3 Flash | 0.969 ± 0.000 | 0.944 ± 0.025 | 0.290 ± 0.032 | 0.310 ± 0.088 | 0.300 ± 0.000 | 0.240 ± 0.052 |
| Column Match - Direct | ||||||
| Claude Opus 4.6 | 0.797 ± 0.016 | 0.787 ± 0.014 | 0.640 ± 0.052 | 0.600 ± 0.071 | 0.320 ± 0.042 | 0.300 ± 0.000 |
| Gemini 2.5 Pro | 0.797 ± 0.016 | 0.831 ± 0.017 | 0.660 ± 0.084 | 0.560 ± 0.089 | 0.180 ± 0.079 | 0.220 ± 0.045 |
| Gemini 3 Flash | 0.803 ± 0.015 | 0.806 ± 0.020 | 0.800 ± 0.000 | 0.710 ± 0.057 | 0.400 ± 0.000 | 0.420 ± 0.042 |
| Column Match - Step-by-Step | ||||||
| Claude Opus 4.6 | 0.938 ± 0.000 | 0.938 ± 0.000 | 0.650 ± 0.053 | 0.660 ± 0.126 | 0.480 ± 0.042 | 0.460 ± 0.052 |
| Gemini 2.5 Pro | 0.884 ± 0.015 | 0.869 ± 0.026 | 0.750 ± 0.071 | 0.780 ± 0.084 | 0.290 ± 0.057 | 0.260 ± 0.055 |
| Gemini 3 Flash | 0.938 ± 0.000 | 0.938 ± 0.000 | 0.800 ± 0.000 | 0.830 ± 0.095 | 0.400 ± 0.000 | 0.420 ± 0.042 |
Notes. Best results per column in each of the four sections of the table are bolded.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.