Kushindwa kwa kategoria za AI BENCHY
Akili ya jumla: Hakufuata maelekezo
Akili ya jumla
Hakufuata maelekezo
Ona ni modeli gani za AI zina uwezekano mkubwa wa kupata Hakufuata maelekezo katika Akili ya jumla, ili uone udhaifu haraka.
Sababu za kushindwa
| Nafasi | Modeli | Kampuni | Idadi ya Hakufuata maelekezo | Alama ya kategoria | Majaribio sahihi | Muda wa majibu (wastani) |
|---|---|---|---|---|---|---|
| #15 | GPT-5.3-Codex medium | OpenAI | 1 | 4.6 | 0/1 | 4.87s |
| #17 | GLM 5 medium | Z.ai | 1 | 6.1 | 0/1 | 14.7s |
| #19 | Seed-2.0-Lite medium | Bytedance Seed | 1 | 6.7 | 0/1 | 18.2s |
| #21 | GPT-5.4 medium | OpenAI | 1 | 4.7 | 0/1 | 4.92s |
| #24 | GPT-5.2 Chat none | OpenAI | 1 | 4.4 | 0/1 | 3.20s |
| #26 | Qwen3.6 Plus medium | Qwen | 1 | 5.1 | 0/1 | 27.1s |
| #28 | Gemini 2.5 Flash medium | 1 | 4.8 | 0/1 | 4.86s | |
| #30 | Qwen3.5-27B medium | Qwen | 1 | 6.1 | 0/1 | 101.4s |
| #31 | DeepSeek V4 Flash high | DeepSeek | 1 | 6.1 | 0/1 | 25.2s |
| #38 | Grok 4.3 medium | X AI | 1 | 5.4 | 0/1 | 24.7s |
| #42 | GPT-5.2 medium | OpenAI | 1 | 3.7 | 0/1 | 4.32s |
| #43 | MiMo-V2.5-Pro medium | Xiaomi | 1 | 5.5 | 0/1 | 4.02s |
| #45 | GPT-5.4 Mini medium | OpenAI | 1 | 4.5 | 0/1 | 3.72s |
| #46 | Qwen3.6 35B A3B medium | Qwen | 1 | 4.4 | 0/1 | 8.66s |
| #49 | Qwen3.5-Flash medium | Qwen | 1 | 6.1 | 0/1 | 40.1s |