Humor Arena
Which frontier model is funniest?
Get picks like this daily. Today's launch numbers, the top launches, AI and tech news, and a daily opportunity pick, straight to your inbox.
Want to reach these readers? Sponsor an issue.
What if you could measure humor?Well we've trained a model on our own dataset of ~50k human ratings to detect what jokes people find funniest. We know it's part objective, part subjective component. Subjective is out of our depth for now hahaThe main results: Fable 5 is funniest - beating the average model 67% of the time, with GPT 4o last at 17%.Other findings: - The models never refused to try, even with dark prompts - Thinking longer has a slight benefit - Absurdness correlates negatively with joke qualitySome methodology notes: - We benchmarked our model against the human majority and it agreed 72% of the time in a blind sample test. - We had 51 US adults rate the jokes, each blind to the models, with joke order randomized, and quality checked for attention and speed. - To rate some yourself visit https://pair.laugh.soThe full benchmark here:https://laugh.so/benchmarkAm taking requests if there's more research you want to see! Cheers
HN · launched Aug 5, 2026 · 10 upvotes (top 42%) · 0 views
Niche: llm benchmarking tools · 84 launches, 1 in the last 7 days · Explore this niche →
Category and tags
- Horizontal
- Observability & eval
- B2C
- AI-native
- Hobby / open-source project
Visit website ↗ Source listing ↗
More like this: all 84 launches in llm benchmarking tools →
Could you build this?
Partial While the web benchmark dashboard is standard, the core value relies on an internal dataset of 50k human humor ratings and an extensively tuned evaluation model.
What it would actually take: The architecture involves an LLM orchestration layer running prompts through dozens of target models, followed by a fine-tuned reward/judge model scoring pairwise humor matchups. Building this requires acquiring proprietary datasets of human humor preference judgments, calibrating Bradley-Terry or Elo rating algorithms, and validating judge correlations against human panels.
Discussion
7 comments analyzed.
Competitors mentioned: OpenCV, Mistral
Concerns raised: Humor assessment is highly subjective and relative, Most LLM responses aren't actually funny, 10x inference cost increase, Difficulty measuring laughter/genuine humor response
Feature requests: Multi-language benchmark dimension, Measure which LLMs succeed at unintentional humor, Test non-US built models performance
Competitors
Other products that read as similar to this one : 78 launches clear the similarity bar, closest 8 shown.
Attention rank: #36 of 79 (itself plus its competitors, highest first, normalized so YC and Product Hunt are compared fairly).
Launched 265 days after the earliest competitor.
- Selbstbild · hn · 2026-07-05 · 6 upvotes · similarity 0.47
- Claude Fable 5.2 and Mythos 5.2 · ph · 2026-11-10 · 2 upvotes · similarity 0.44
- Funny Guys · ph · 2026-09-18 · 2 upvotes · similarity 0.44
- awesome-fable5-5-videos · github · 2026-10-05 · 41 upvotes · similarity 0.43
- Echo · hn · 2026-07-23 · 484 upvotes · similarity 0.43
- GiggleTime: Jokes & Comedians · ph · 2026-10-18 · 1 upvotes · similarity 0.43
- awesome-fable5-2-videos · github · 2026-11-10 · 11 upvotes · similarity 0.41
- Antics: Drop-in multiplayer for your AI-built games · hn · 2026-07-29 · 5 upvotes · similarity 0.40
The niche, labels, vibe code rating and competitor comparison on this page are analysis by Nicheloom. If you use them, please credit nicheloom.com. How you may use our data.