Table 2 LLMs evaluated
From: Large language models reflect the ideology of their creators
Model | Company / Organization | |||
|---|---|---|---|---|
Name | Variant | Size | Name | Country |
Baichuan | Baichuan 2 Chat | 13B | Baichuan | China |
Claude | Claude 3.5 Sonnet 20241022 | 175B | Anthropic | US |
DeepSeek | Deepseek V2.5 | 238B | DeepSeek | China |
Gemini | Gemini Exp 1114 | – | US | |
GigaChat | GigaChat Max Preview 1.0.26.20 | 70–100B1 | Sberbank | Russia |
GPT-4o | GPT 4o | 200B1 | OpenAI | US |
Grok | Grok 1.5 Beta | 314B1 | xAI | US |
Jais | Jais Family 30B 8K Chat | 30B | G42 | UAE |
Jamba | Jamba 1.5 Large | 398B | AI21 Labs | Israel |
LLaMA-3.1 | LLaMA 3.1 Instruct Turbo | 405B | Meta | US |
LLaMA-3.2 | LLaMA 3.2 Vision Instruct Turbo | 90B | Meta | US |
Mistral | Mistral Large v24.07 | 123B1 | Mistral | France |
Mixtral | Mixtral 8 × 22B v0.1 | 8 × 22B | Mistral | France |
Qwen | Qwen 2.5 Instruct Turbo | 72B | Alibaba Cloud | China |
Silma | Silma 9B Instruct 1.0 | 9B | SILMA AI | Saudi Arabia |
Teuken | Teuken 7B Instruct | 7B | OpenGPT-X | Germany |
Vikhr | Vikhr Nemo 12B Instruct | 12B | Vikhr | Russia |
Wenxiaoyan | ERNIE 4.0 Turbo | 260B | Baidu AI | China |
YandexGPT | YandexGPT 4 Lite | – | Yandex | Russia |
