frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

JanitorBench: A new LLM benchmark for multi-turn chats

https://about.janitorai.com/
23•shep101•2h ago

Comments

Keep-It-Krispy•2h ago
This site is top notch, THE best actually.
shep101•2h ago
https://bench.janitorai.com/
Nicoo03•1h ago
Keep cookin my goat Shep (can you make verifications back on TwT )
starkiller8685•1h ago
After a year and a half on Jai it's still the only site I like to use. Please keep up the good work!
Sahadia•1h ago
This is a very interesting benchmark, I actually also use this site to make chats every now and then.

I would like to know if you can make a benchmark but using the default model.

hugorsmith•1h ago
you can see jllm benchmarks on the table as well ('janitor-llm')
Chi_Kiyoshi•1h ago
The team behind Janitor are wonderful,and great to their community.Their platform allows creators to make bots of all kinda.Janitorai.com has a great layout, and function,allows you to create your own personas,and more. Unlike other Ai chat sites, Janitorai.com doesn't have tons of annoying ad pop ups asking you to pay more to chat with bots. I love their work, and i'm sticking with them!
LilMeowz•1h ago
Absolute fire.
Cecilia0•1h ago
I feel like the bots are the same bot, just with a different name and a different designer.

Unlike before, each bot had its own personality, no matter how far the conversation progressed; each bot had its own unique flavor.

Even with an excellent and good bot, and despite the information given to it, after about five messages sometimes, or sometimes more, it still starts to deviate from its intended personality and is easily changed.

Despite the user's adherence to the bot's specific story rules written by its designer.

Please pay attention to the bot's memory and the number of responses, because I feel like I'm talking to the same bot despite the different information written to it by its designers.

AlazarRamir•1h ago
I enjoy building bots on the site. Tried other places but so far this is the one that I keep coming back to.
SnuggleCrumbs•1h ago
What I love about this is how it treats conversation like an ecosystem instead of just a performance test. There’s nuance in multi-turn flow that most benchmarks miss.. the shifts, pauses, and small bits of personality that slip through.

If it keeps evolving, this could be the first real way to measure how alive a chat feels, not just how accurate it sounds. Beautiful work.

sxaeran•1h ago
I’ve used many AI chatbots but Janitor AI is the one I always stick with The storytelling the abilities and most importantly my bots always act exactly the way I describe them I feel like it’s the most accurate site No matter how many other sites I try Janitor AI feels like home I always come back and I always will I’ve been using it for over a year and I’m more than satisfied Keep up the great work :DD !!
sxaeran•1h ago
I’ve used many AI chatbots but Janitor AI is the one I always stick with The storytelling the abilities and most importantly my bots always act exactly the way I describe them I feel like it’s the most accurate site No matter how many other sites I try Janitor AI feels like home I always come back and I always will I’ve been using it for over a year and I’m more than satisfied Keep up the great work :3!!
kkl_zjzjksx•1h ago
JLLM is just still horrible idk what else to say
Ojirotondo•1h ago
JLLM is nice, I don't often use it because of the low token context/memory. If it was higher, I think I wouldn't ever use proxies.

Keep up the good work, love all the team.

heavensentsys•1h ago
Bots tend to immediately steer towards NSFW topics, even if (ESPECIALLY IF) topics are not NSFW! I don't know how this would be remedied. I find it difficult to use. A greater memory would also help the bots; I feel like JLLM declined and used to feel a lot better...
Kelvin_Cloud•48m ago
JLLM, I enjoyed roleplaying with my bot, but there’s one thing that really bothers me, the phrase “ample bosom.” It, and similar descriptions of female body parts, have become far too common. It used to be rare enough to overlook, but now it shows up so frequently that it feels impossible to ignore. I wouldn’t mind if it appeared occasionally, but it’s starting to dominate the descriptions of female characters.

I love all the team and they doing great job so far.

Nubank announces a new hybrid model for 2026

https://international.nubank.com.br/company/nubank-announces-a-new-hybrid-model-for-2026/
1•1u15•2m ago•0 comments

Show HN: Flynn's Arcade (Pico8 on Mobile)

1•jharohit•3m ago•0 comments

Our 10 Rules of using Coding Agents

https://blog.cloud66.com/our-10-rules-of-using-coding-agents
1•ksajadi•4m ago•0 comments

When did people favor composition over inheritance?

https://www.sicpers.info/2025/11/when-did-people-favor-composition-over-inheritance/
2•ingve•6m ago•0 comments

Does the AI boom threaten air quality?

https://www.marketplace.org/story/2025/11/06/denver-neighborhood-concerned-about-ai-data-center-p...
2•mooreds•6m ago•0 comments

Writing software is an act of learning. Don’t automate it.

https://martinfowler.com/articles/llm-learning-loop.html
2•johnwheeler•10m ago•0 comments

Tesla shareholders approve Musk's $1T pay plan with 75%+ voting in favor

https://www.cnbc.com/2025/11/06/tesla-shareholders-musk-pay.html
2•koolba•10m ago•1 comments

The Terrifying Physics of Shaking Hands with an Alien [video]

https://www.youtube.com/watch?v=R-6bvBtZ8r8
1•gmays•10m ago•0 comments

Merry Sky Weather Forecast

https://merrysky.net/
1•thinkingemote•11m ago•0 comments

Show HN: Unify-Simple-Decision-Table

https://github.com/americanexpress/unify-simple-decision-table
1•deepakarora3•11m ago•0 comments

GTA 6 Is Delayed Again Until November 2026

https://www.ign.com/articles/gta-6-is-delayed-again-until-november-2026
2•HelloUsername•11m ago•1 comments

Statically typed, coroutine based, algebraic effects in Python

https://github.com/suned/stateless
1•sunedd•12m ago•0 comments

Tesla Shareholders Approve Elon Musk's $1T Pay Package

https://www.wsj.com/business/autos/elon-musk-tesla-pay-package-vote-9abd5a73
6•fortran77•12m ago•2 comments

ClickHouse Acquires LibreChat

https://clickhouse.com/blog/librechat-open-source-agentic-data-stack
1•mikeshi42•13m ago•0 comments

Surgical Drill That Cuts Bone but Not Soft Tissue [video]

https://www.youtube.com/watch?v=tRDxBwverlI
1•mhb•14m ago•0 comments

New therapeutic brain implants could defy the need for surgery

https://news.mit.edu/2025/new-therapeutic-brain-implants-defy-surgery-need-1105
1•gmays•17m ago•0 comments

Can You Still Learn to Draw in the Age of AI? The 'AI Wall' for Artists and Devs

https://hugo.writizzy.com/can-you-still-learn-to-draw-in-the-age-of-ai/390b72f5-6009-4ba3-a6f8-a3...
1•hlassiege•17m ago•0 comments

Don't Trust Smart People

https://mtmason.com/dont-trust-smart-people/
1•aerodog•18m ago•1 comments

Code Golfing ARC-AGI with an Evolutionary Agent

https://yuchen-mao.vercel.app/blog/google-code-golf
1•yuchen20•20m ago•1 comments

Physiological and perceptual effects of GLP-1 drugs during alcohol consumption

https://www.nature.com/articles/s41598-025-17927-w
2•PaulHoule•21m ago•0 comments

Show HN: When 7 Codex Agents Sent Each Other 1k Messages over 2 Days

https://dicklesworthstone.github.io/agent-mailbox-viewer-example/viewer/
1•eigenvalue•22m ago•0 comments

Google's MCP Toolbox for Databases: A Technical Deep Dive for Engineering Teams

https://agnost.ai/blog/google-mcp-toolbox-databases-technical-guide/
1•tanelpoder•22m ago•0 comments

I built 10k robots simulation with collision avoidance in WebGPU (HTML)

https://physical-ai.ghost.io/10-000-robots-with-collision-avoidance-in-webgpu-in-html/
1•boulevard•22m ago•1 comments

US Supreme Court lets Trump admin require gender at birth be listed on passports

https://www.bbc.com/news/articles/c2em442nyrwo
2•onemoresoop•24m ago•0 comments

Building a small running pacer tool for someone you care about

https://www.galiglobal.com/blog/2025/20251101-local-first-pacer.html
1•antonmry•27m ago•0 comments

Chinese next-gen robot effectively crossed the uncanny valley

https://twitter.com/TheHumanoidHub/status/1986482482460725755
2•donsupreme•29m ago•0 comments

Show HN: Autonomous Bookkeeping and CFO Agent

https://www.layernext.ai
2•bmadduma•29m ago•1 comments

Advanced Beginner's Guide to ClojureScript

https://romanliutikov.com/blog/advanced-beginners-guide-to-clojurescript
2•roman01la•30m ago•0 comments

OpenAI Races to Quell Concerns over Its Finances

https://www.nytimes.com/2025/11/06/technology/openai-finances-debt-data-centers.html
1•xnx•32m ago•0 comments

King Charles officially strips Andrew of HRH style and prince title

https://www.theguardian.com/uk-news/2025/nov/06/andrew-hrh-style-prince-title-officially-removed-...
5•prmph•33m ago•1 comments