AI BENCHY Fouten
Verkeerd antwoord-fouten
Zie welke AI-modellen het vaakst tegen Verkeerd antwoord aanlopen, zodat je betrouwbaarheidsrisico's ziet voordat je kiest. Sorteren op: Aantal fouten ↑.
Categorieën
| Rang | Model | Bedrijf | Verkeerd antwoord-aantal | Score | Correcte tests | Responstijd (gem.) |
|---|---|---|---|---|---|---|
| #98 | LFM2-24B-A2B none | Liquid | 9 | 4.1 | 1/16 | 811ms |
| #61 | Seed-2.0-Lite none | Bytedance Seed | 10 | 6.2 | 8/18 | 2.53s |
| #62 | Gemini 2.5 Flash none | 10 | 6.2 | 7/18 | 903ms | |
| #66 | GPT-5.4 none | OpenAI | 10 | 5.9 | 7/18 | 1.51s |
| #67 | Qwen3.5-27B none | Qwen | 10 | 5.9 | 6/18 | 1.74s |
| #74 | GLM 4.7 Flash none | Z.ai | 10 | 5.6 | 5/18 | 3.35s |
| #75 | GLM 5.1 none | Z.ai | 10 | 5.6 | 5/18 | 4.33s |
| #77 | GLM 5 Turbo none | Z.ai | 10 | 5.5 | 6/18 | 2.94s |
| #79 | Grok 4.20 Beta none | X AI | 10 | 5.3 | 4/18 | 1.19s |
| #86 | GPT-5.4 Mini none | OpenAI | 10 | 5.1 | 5/18 | 1.17s |
| #88 | Nemotron 3 Super none | NVIDIA | 10 | 5.1 | 4/18 | 8.54s |
| #90 | Qwen3.5-9B none | Qwen | 10 | 4.8 | 4/18 | 1.47s |
| #70 | Qwen3.5-122B-A10B none | Qwen | 11 | 5.7 | 6/18 | 3.69s |
| #78 | Trinity Large Preview none | Arcee AI | 11 | 5.3 | 5/18 | 5.07s |
| #83 | Mistral Small 4 none | Mistral | 11 | 5.2 | 5/18 | 665ms |