Quick Setup Guide: (Free!) OpenRouter Setup
Formamorph » Devlog
1. Create OpenRouter Account & Get API Key
- Go to https://openrouter.ai
- Create an account
- Generate an API key from your dashboard
2. Choose Your Model
- Recommended free model: mistralai/mistral-7b-instruct:free
- Copy the exact model name
3. Configure Game Settings
- Open game settings (top right menu button in-game)
- Change API URL to: https://openrouter.ai/api/v1/chat/completions
- Paste your API key
- Enter your chosen model name
- optional: In the same setting menu, increase memory limit from 2k to 3k if the model’s max memory is 5k (just subtract 1-2k from the model’s max context length)
That’s it! You should now be able to play the game using faster AI models!
4. So…
- There’s a limit to small free models, however. The little 7b model will likely fail when the world is too big and complex like Veilwood :(
- If you want to play in a really large complex world, I recommend spending a few bucks on openrouter and use a paid model such as mistral nemo instruct. It doesn’t have to be an expensive model, just one that is smart and creative.
Important:
- Free models have rate limits!
- Different models have more/less limits. You can experiment by changing free models.
Best free model: meta-llama/llama-3.3-70b-instruct:free
- This model is very smart and has extreme memory (128K!)
- HOWEVER: it has more limited rate limits which means you can’t send in many requests, maybe once per minute
Files
dist.zip Play in browser
40 days ago
Get Formamorph
Formamorph
Every choice transforms your body and shapes your adventure
| Status | In development |
| Author | FieryLion |
| Genre | Interactive Fiction |
| Tags | Adult, bbw, belly, Erotic, Female Protagonist, inflation, pregnancy, vore, weight-gain |
More posts
- v 1.1.0 & v1.1.3 Major Improvements!32 days ago
- v1.0.1 & v1.0.2 & v1.0.3 patch38 days ago
- Launch! v1.038 days ago
Comments
Log in with itch.io to leave a comment.
Does this work with the NovelAI API or is it only geared toward OpenRouter's library? OpenRouter has enough models on its own, for sure. I just have a NovelAI subscription already and I'm wondering if I can leverage it.
Does the meta-llama/llama-3.3-70b-instruct:free model have a limit? It stopped working a few hours of use.
I think it may have a daily limit
I can’t seem to find mistral Nemo instruct on open router. Only mistral Nemo. Mistral Nemo is the one I’m using and it seems inconsistent. What other one could I use that would be similar to mistral Nemo instruct?
I assume you’re looking for paid models, then mistralai/mixtral-8x7b-instruct is very smart and uncensored
thank you!
with that AI is there a way to make it text a bit more? in the settings i have max memory at 33k and max output tokens at 8k but it seems to only make around 3-5 sentences.
in settings, uncheck the Single Paragraph Event option
thank you so much! sorry I have been posting comments so much. I really apricate your help :)!
you’re welcome!
i'm getting a ton of error messages saying "the ai model was unable to produce the correct json format. try a different model." and i thought i followed the tutorial perfectly, sometimes it works and i just have to wait a bit to get to the next hour but it's a bit annoying, sometimes the default ai does this too
![]()
That means you probably need to use a smarter, bigger model (make sure the model is an instruct model). The default AI isn’t great either sorry but my potato PC can’t run the good stuff.
what model do you recommend apart from 7b that is smarter
anything larger that is an instruct model, I strongly recommend nemo, or if you are paying for an AI provider like open router then Magnum 72B or bigger llama finetunes
is there a way to run this locally?
if you know how to setup local LLM, just use your localhost URL and port number instead of the openrouter URL.
However you have to use the download version because you cannot enter localhost URL into the itch.io browser game due to security policies
The easiest way is through Ollama. Download and install it. After that, open a command prompt and type in ollama run mistral:7b-instruct
After the installation is done and the model is running, open the game and in the configurations, for Endpoint URL, type in http://localhost:11434/v1/chat/completions and to Model name, type in mistral:7b-instruct
After that you should be ready. I must say tho, that is is not working that well, mostly because the model is dumb. If you can run a better model, you can do it with Ollama, you will just need to host the model with a different command and set the model name to the name of the model you chose.
(edit: if you encounter issues like game out of VRAM, try to lower the Max Memory in settings)