Eșecuri AI BENCHY
Eșecuri Apel de instrument invalid
Vezi ce modele AI se lovesc cel mai des de Apel de instrument invalid, ca să identifici riscurile de fiabilitate înainte să alegi.
| Rang | Model | Companie | Număr de Apel de instrument invalid | Scor | Teste corecte | Timp de răspuns (mediu) |
|---|---|---|---|---|---|---|
| #136 | Elephant Alpha medium | Openrouter | 1 | 5.1 | 6/21 | 1.27s |
| #137 | Elephant Alpha none | Openrouter | 1 | 5.1 | 5/21 | 1.22s |
| #139 | DeepSeek V4 Flash none | DeepSeek | 1 | 5.0 | 5/21 | 26.8s |
| #145 | Laguna M.1 none | Poolside | 1 | 4.8 | 4/19 | 2.89s |
| #146 | Laguna Xs.2 none | Poolside | 1 | 4.8 | 5/19 | 806ms |
| #154 | Qwen3.5-9B none | Qwen | 1 | 4.6 | 4/21 | 1.89s |
| #158 | GLM 4.7 Flash medium | Z.ai | 1 | 4.4 | 4/21 | 35.1s |
| #159 | Ling-2.6-1T none | Inclusionai | 1 | 4.3 | 3/21 | 7.72s |
| #163 | Granite 4.1 8B none | IBM Granite | 1 | 4.0 | 2/21 | 728ms |