Benchmark Almanac

BuzzBench

Gemini 2.5 Pro Mar 2025 71.1 leads since March 2025
Not in model cards

Models analyze humor, such as why a joke or comic piece works. The leaderboard reports a score and response length per model.

The site calls it a humor analysis benchmark. The grading method was not read, so scoring type needs a check.

Kind
Taste and style
Level
Everyday
Scored
Graded by a model
Answers
Open-ended
Tasks
All public
Strongest model tested
Gemini 2.5 Pro Mar 2025 #103 on the capability index
Last model added
April 2025
Scores refreshed
November 12th, 2026
26
Entries on its own leaderboard, 20 models we track

Gemini 2.5 Pro leads with 71.1 among models released by April 2025

Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

The newest model with a score here came out in April 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

26 entries. Scroll sideways for the rest.

  1. 71.1 Gemini 2.5 Pro
  2. 69.1 GPT-4o
  3. 68.2 Claude 3.7 Sonnet
  4. 67.8 quasar-alpha
  5. 67.1 GPT-4.1
  6. 62.7 deepseek-ai/deepseek-r1 (snapshot unspecified)
  7. 61.9 claude-3.5-sonnet (snapshot unspecified)
  8. 59.5 chatgpt-4o-latest
  9. 59.3 o1
  10. 54.9 gemini-pro-1.5 (snapshot unspecified)
  11. 54.4 Gemini 2.0 Flash
  12. 53.3 Claude 3.5 Haiku
  13. 51.8 Grok-2
  14. 51.6 deepseek-ai/DeepSeek-V3 (snapshot unspecified)
  15. 47.5 Mistral Large 2
  16. 46.2 o1-mini
  17. 45.9 Llama 3.1-405B
  18. 45.9 Qwen2.5-72B
  19. 45.7 Llama 3.3 70B Instruct
  20. 39.0 GPT-3.5 Turbo
  21. 38.7 Phi-4
  22. 32.9 Llama 3.1-8B
  23. 32.3 Ministral 8B
  24. 31.4 Qwen2.5-7B
  25. 26.6 Llama 3.2 3B
  26. 9.51 Llama 3.2 1B
  1. 1 Gemini 2.5 Pro Mar 2025 71.1
  2. 2 GPT-4o Mar 2025 69.1
  3. 3 Claude 3.7 Sonnet 68.2
  4. 4 quasar-alpha 67.8
  5. 5 GPT-4.1 67.1
  6. 6 deepseek-ai/deepseek-r1 (snapshot unspecified) 62.7
  7. 7 claude-3.5-sonnet (snapshot unspecified) 61.9
  8. 8 chatgpt-4o-latest 59.5
  9. 9 o1 59.3
  10. 10 gemini-pro-1.5 (snapshot unspecified) 54.9
  11. 11 Gemini 2.0 Flash Feb 2025 54.4
  12. 12 Claude 3.5 Haiku 53.3
  13. 13 Grok-2 Dec 2024 51.8
  14. 14 deepseek-ai/DeepSeek-V3 (snapshot unspecified) 51.6
  15. 15 Mistral Large 2 Nov 2024 47.5
  16. 16 o1-mini 46.2
  17. 17 Llama 3.1-405B 45.9
  18. 18 Qwen2.5-72B 45.9
  19. 19 Llama 3.3 70B Instruct 45.7
  20. 20 GPT-3.5 Turbo Jun 2023 39.0
  21. 21 Phi-4 38.7
  22. 22 Llama 3.1-8B 32.9
  23. 23 Ministral 8B 32.3
  24. 24 Qwen2.5-7B 31.4
  25. 25 Llama 3.2 3B 26.6
  26. 26 Llama 3.2 1B 9.51

Source: Official BuzzBench leaderboard

#EntryScoreSettingDateSource
1 Gemini 2.5 Pro Mar 2025 71.1 Mar 2025 Official leaderboard
2 GPT-4o Mar 2025 69.1 Mar 2025 Official leaderboard
3 Claude 3.7 Sonnet 68.2 Feb 2025 Official leaderboard
4 quasar-alpha 67.8 Official leaderboard
5 GPT-4.1 67.1 Apr 2025 Official leaderboard
6 deepseek-ai/deepseek-r1 (snapshot unspecified) 62.7 Official leaderboard
7 claude-3.5-sonnet (snapshot unspecified) 61.9 Official leaderboard
8 chatgpt-4o-latest 59.5 Official leaderboard
9 o1 59.3 Dec 2024 Official leaderboard
10 gemini-pro-1.5 (snapshot unspecified) 54.9 Official leaderboard
11 Gemini 2.0 Flash Feb 2025 54.4 Feb 2025 Official leaderboard
12 Claude 3.5 Haiku 53.3 Oct 2024 Official leaderboard
13 Grok-2 Dec 2024 51.8 Dec 2024 Official leaderboard
14 deepseek-ai/DeepSeek-V3 (snapshot unspecified) 51.6 Official leaderboard
15 Mistral Large 2 Nov 2024 47.5 Nov 2024 Official leaderboard
16 o1-mini 46.2 Sep 2024 Official leaderboard
17 Llama 3.1-405B 45.9 Jul 2024 Official leaderboard
18 Qwen2.5-72B 45.9 Sep 2024 Official leaderboard
19 Llama 3.3 70B Instruct 45.7 Dec 2024 Official leaderboard
20 GPT-3.5 Turbo Jun 2023 39.0 Jun 2023 Official leaderboard
21 Phi-4 38.7 Dec 2024 Official leaderboard
22 Llama 3.1-8B 32.9 Jul 2024 Official leaderboard
23 Ministral 8B 32.3 Oct 2024 Official leaderboard
24 Qwen2.5-7B 31.4 Sep 2024 Official leaderboard
25 Llama 3.2 3B 26.6 Sep 2024 Official leaderboard
26 Llama 3.2 1B 9.51 Sep 2024 Official leaderboard

7 models have held number one on the official leaderboard since 2023

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI Meta Anthropic Google DeepMind
Own leaderboard Not tested since Apr 2025
Gemini 2.5 Pro 71.1

Source: Official BuzzBench leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard
0.00 20.0 40.0 60.0 80.0 2023 2024 2025 0.00 20.0 40.0 60.0 80.0 2023 2024 2025

Source: Official BuzzBench leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Writing and taste

All 30 in Writing and taste
  • Design Bench (Moonshot) In use Models make graphic designs from direct prompts or few-shot reference images. An automated grader rates the results as Elo scores. Kimi K3.1 Pro 957
  • WritingBench Fading Models complete 1,000 real-world writing requests across 6 domains and 100 subdomains. A judge model scores each reply on five query-specific criteria. GPT-5 83.9
  • Creative Writing (MiniMax in-house) Only in system cards Models write creative pieces for real Hailuo AI user requests. Automatic assessments score the writing. MiniMax-Text-01 81.3%
  • DeepSeek Chinese functional writing evaluation Only in system cards Models answer 3,170 everyday Chinese writing prompts such as reports, emails and speeches. Pairwise judging against a baseline model gives a win rate. DeepSeek-V4-Pro 62.6%
  • Slide generation human evaluation (Z.ai) Only in system cards People compare HTML slide decks from two models on content, layout and visual quality. Metric is overall win rate. GLM-5 67.5%
  • Task-dependent output homogenization evaluation Only in system cards Models answer prompts several times. A judge scores functional diversity where it is wanted and consistency where one answer is right. Muse Spark 0.21