Kushindwa kwa kategoria za AI BENCHY
Ufuataji wa maagizo: Hakufuata maelekezo
Ufuataji wa maagizo
Hakufuata maelekezo
Ona ni modeli gani za AI zina uwezekano mkubwa wa kupata Hakufuata maelekezo katika Ufuataji wa maagizo, ili uone udhaifu haraka. Panga kwa: Idadi ya kushindwa ↑.
Modeli zilizoonyeshwa
15
Jumla ya kushindwa
22
Modeli iliyoathirika zaidi
Gemini 3.1 Flash Lite Preview 1Sababu za kushindwa
| Nafasi | Modeli | Kampuni | Idadi ya Hakufuata maelekezo | Alama ya kategoria | Majaribio sahihi | Muda wa majibu (wastani) |
|---|---|---|---|---|---|---|
| #11 | Gemini 3.1 Flash Lite Preview high | 1 | 7.9 | 1/2 | 70.1s | |
| #25 | Grok 4.20 Beta medium | X AI | 1 | 8.3 | 1/2 | 4.97s |
| #30 | Step 3.5 Flash medium | Stepfun | 1 | 8.5 | 1/2 | 4.98s |
| #35 | MiMo-V2-Omni medium | Xiaomi | 1 | 8.3 | 1/2 | 4.92s |
| #44 | GPT-5.4 Mini medium | OpenAI | 1 | 7.4 | 1/2 | 2.50s |
| #45 | GPT-5 Mini medium | OpenAI | 1 | 8.0 | 1/2 | 15.7s |
| #51 | Nemotron 3 Super medium | NVIDIA | 1 | 7.2 | 1/2 | 7.72s |
| #52 | Grok 4.1 Fast medium | X AI | 1 | 6.6 | 1/2 | 5.30s |
| #56 | Grok 4.20 Multi Agent Beta medium | X AI | 1 | 8.3 | 1/2 | 4.63s |
| #57 | GPT-5 Nano medium | OpenAI | 1 | 8.5 | 1/2 | 11.9s |
| #60 | Gemma 4 26B A4B none | 1 | 4.4 | 0/2 | 1.08s | |
| #71 | MiniMax M2.5 medium | Minimax | 1 | 8.1 | 1/2 | 4.64s |
| #75 | GLM 5.1 none | Z.ai | 1 | 8.3 | 1/2 | 1.58s |
| #78 | Trinity Large Preview none | Arcee AI | 1 | 4.1 | 0/2 | 1.09s |
| #79 | Grok 4.20 Beta none | X AI | 1 | 4.8 | 0/2 | 687ms |