19
Why not consider one of the dozens of solo ttrpg systems that exist instead of cramming DND into a niche it's not designed for?
Also LLMs aren't "they" any more than they're "he" or "she"
Short answer: no.
Longer answer: noooooooooooo.
Google AI Studio has a massive token window. I've tested it out using a setup where I feed a templated prompt into Gemini 2.5 pro along with a PDF of any particular campaign (courtesy of Anna's Archive), and the campaign actually plays out really well.
Here's the prompt and a readme I put into google drive Just be sure to change the {character name/race/class} for your character in the prompt before you submit it.
Since it's a large model, a 50 response session might have an equivalent energy cost of running a microwave for 7 minutes (or more -- and of course that doesn't count the energy used in the training of the model as well), so make of that what you will.
eating my kraft dinner raw so i can run a game om the word calculator
Do you have any gameplay examples resulting from this?
What if you just play Gloomhaven?
If you want a solo DnD you should understand that you're not getting DnD but at best a co-written story in second person perspective
Look into SillyTavern and Worldlore in the docs
Understand that this will actually take hours or even days to set up, and in the end, it'll still be an AI that is ass at its job, something you have to fix every few messages
You should also know that self-hosting ai may not be enough, 4k tokens is bare minimum and even that usually takes up 8gb of vram, DeepSeek API is cheap, use that.
Get fucking smashed and put all your psychic energy into creating a D&D tulpa
Maybe try to look for a solo adventure book? DND isn't the best for this, but you could do it. Or just play baldur's gate 3
For a local chatbot I guess you could download Ollama and then the various models via Ollama on your system (it has DeepSeek and Qwen). Ollama will run the models locally. Then you can use either Chatbox or Msty to serve as a front-end for your llm models. Msty has something called 'Knowledge Stacks' where you can do things like load your core rulebooks and maybe pdfs or epubs of other stories you want inspiration from into a database that your llm models can reference across different conversations. Chatbox has something called Knowledge Bases that can allegedly do the same thing but I could not get that to work.
No, I would suggest not. I tried to run a solo campaign with chatgpt early on out of curiousity but it fundamentally 1) didn't understand the dnd system and would never prompt for challenge rolls unless frequently reminded, 2) didn't understand what difficulty the roll should be if it did. Then thirdly it just didn't push back at all, because of the agreeableness factor anything I would 'attempt' would always succeed. It also didn't have any story in mind it was just lole like pushing at an open door every time into another scene of the most mundane fantasy story imaginable. 1/10 do not recommend
Ah, thanks; microsoft's copilot is very similar in its issues; it can actually store your and an enemy's hp values and is actually capable of random rolls, but it's also ridiculously agreeable, such as allowing me to directly tell an enemy that I'm leaving to gather a party to come back and kill it and it just lets me walk away.
the environmental destruction llms cause means it's never appropriate to use them for anything
I tend to agree regarding the environmental angle to abolish llms, but in cases like this I'm not sure that's a fair comparison. I don't have the resource usage numbers, but if I had to venture a guess regarding environment cost the actual cost of using an llm for something like this vs playing baldurs gate 3 for a similar experience, surely bg3 would blow the llm usage (not training) costs out of the water. The environmental cost of buying a printed DnD rulebook would also by itself be more.
Granted, a user could just find a PDF or whatever, but thats not really my point in this case. I just think we probably need to have concrete understanding of the costs of llms (and where it comes from) when levying critiques of the technology.
regarding environment cost the actual cost of using an llm for something like this vs playing baldurs gate 3 for a similar experience, surely bg3 would blow the llm usage (not training) costs out of the water.
Unfortunately you're way off. Running an LM locally will be slower and less accurate than using an LLM, and still blow bg3 out of the water for power usage due to it needing everything the graphics card has in order to process anything. Processing natural language is massively more intensive than just displaying some models and pretty effects. Move to an LLM and you're consuming even more power per query even faster as it accesses more graphics cards. Add in training and the disparity gets even bigger.
I'm cool with being wrong about this, but unfortunately haven't found a solid source recently to indicate either way (maybe another day I'll have time to find one, but not today)
Unfortunately my comparison was obv apples to oranges, but id find it at least a little bit surprising if a continuously running and rendering process like a game would use the same amount of energy in the same amount of time as a user interacting with a chatbot that only uses a GPU on request. Yes, if a user was continuously feeding input in and generating output from a llm that seems trivially true that it would use more energy, but thats not how it would be used in this case.
Also, training is a fair point for end-to-end analysis but I don't even know how we would start that for total energy cost for developing a game, hence leaving that out of discussion. (Edit: I assume, because of all the attention on llms in recent years, that training an llm costs significantly more than developing a game (how ever we would fence off both measurements), but that is even more vibes based analysis than my original assumption earlier in the thread)
Yeah, the time spent between responses does reduce the energy consumption, but games don't use 100% of the GPU all the time either - they'll raise or lower the power they're pulling in response to how intensive the graphics currently are, and can be bottlenecked by the CPU. BG3 is kinda of a bad example game because despite the polish it's a CRPG, so most of the processing is directed towards NPC models and behaviours, which is handled by the CPU - the GPU is very unlikely to spin all the way up to max power like it will be the entire time it's processing an LM query.
Oh my God I hate this; I asked deepseek to create for me shadowrun partners and it's embraced the theme so beautifully; the hacker 0xMiraj has a keffiyeh with embedded fiber optics cables and has as his contact an AI imam of a hacktivist mosque called Al muezzin (apparently the 'muezzin' is the person who conducts the Muslim call to prayers), a rigger called ghosthaul who's a Bedouin smuggler with a drone army, and a physical adept called al razi who has as his contact an underground cyber-monk network called the blind sheikh.
Ah I really hate this; I don't want to lose all this, especially when Im quite certain the chatGPT, copilot and grok versions will just be racist as heck.
This isn't related to your question, but you wrote "apparently the 'muezzin' is the person who conducts the Muslim call to prayers" and I have to wonder: how have you lived in Qatar for over a decade and you're just now learning what a muezzin is?
I thought the imam was the one who did that (I don't really know why the two are separate roles)
I'd just look for a free ttrpg on ~~twitch~~ itch.io made for one player, there are a bunch of those.
I'm interested in this, any good ones or general idea how I'd go about finding one?
I said twitch but I meant itch.io. I don't actually know much about them, I just have spent many hours trying to find obscure RPGs (mostly of the JRPG style) and the filtering on itch isn't fantastic. You can find a ton though, since I found several when I wasn't even looking for them.
Got it and no worries thought it be like a interactive thing, I actually got a bunch I need to sort through that I got in bundles before
Me too
Run a custom model on your own hardware. DavidAU makes models specifically tuned for story telling
Have you tried a GM Emulator Oracle? Something like: https://www.wordmillgames.com/page/mythic-gme.html
You could always try to set up a group here, I actually put together a shadowrun group from hexbear users and had a successful 1 year campaign that went through many of the 1e-2e adventure modules
Download LM studio and use Qwen or llama
this post was submitted on 22 Jul 2025
19 points (91.3% liked)
chapotraphouse
13950 readers
663 users here now
Banned? DM Wmill to appeal.
No anti-nautilism posts. See: Eco-fascism Primer
Slop posts go in c/slop. Don't post low-hanging fruit here.
founded 4 years ago
MODERATORS