BuzzBench

Name: BuzzBench
Published: 2025-01-05

Chat & Writing · 2025-01-05

Humor-analysis benchmark for emotional-intelligence evaluation.

Model	Score
Gemini 2.5 Pro Preview 03-25	71.1
ChatGPT-4o Latest observed 2025-03-27	69.1
Claude 3.7 Sonnet	68.2
GPT-4.1	67.1
DeepSeek-R1	62.7
Claude 3.5 Sonnet (Oct 2024)	61.9
ChatGPT-4o Latest (source-unspecified snapshot)	59.5
o1 (source unspecified)	59.3
Gemini 1.5 Pro (source unspecified)	54.9
Gemini 2.0 Flash	54.4
Haiku 3.5	53.3
Grok 2 1212	51.8
DeepSeek-V3	51.6
Mistral Large 2.1 (Instruct 2411)	47.5
o1-mini	46.2
Llama 3.1 405B Instruct	45.9
Qwen2.5 72B Instruct	45.9
Llama 3.3 70B Instruct	45.7
GPT-3.5 Turbo 0613	39.0
Phi-4	38.7
Llama 3.1 8B Instruct	32.9
Ministral 8B Instruct 2410	32.3
Qwen2.5 7B Instruct	31.4
Llama 3.2 3B Instruct	26.6
Llama 3.2 1B Instruct	9.5