Introducing SissyGPT 2.0 and the LLM Lab
- Alyxsis
- Jul 11
- 4 min read
Greetings, sissies! I bring lots of news. Today we're launching the biggest update yet for SissyGPT, introducing the new LLM Lab, adding the ability to use any custom model in HornyHub, and more improvements site-wide. Let's dive in!
SissyGPT
New Providers
SissyGPT has been entirely remastered with many improvements and has been given a big makeover with much more identity. Most notably, it now supports dozens of top models via OpenRouter, not just models from OpenAI. SissyGPT already used various methods to bypass OpenAI's censorship, but their censors get more sophisticated every week; meanwhile, OpenAI has stagnated and underperformed compared to other top labs. SissyGPT still provides access to OpenAI's models, but you can now choose any model you want to be the brain of your mistress! Utilize models from labs like Google DeepMind, xAI, DeepSeek, Anthropic, Mistral AI, Meta AI, Alibaba AI, and Nous Research. More on this below ~
Reasoning Models
SissyGPT now supports Reasoning models! These models have a thorough thought process that they work through before responding to the user. This means they may take slightly longer to respond, but it allows for much better long-term planning as they "think out loud" to themselves. Instead of simply predicting the most likely next word, they're able to formulate complex plans without cluttering the chat, emulate characters with depth, and self-correct if they (or you 😳) make any mistakes in their roadmap.
Currently, all the smartest models in the world are Reasoning models, such as Gemini-2.5-Pro, DeepSeek-R1, and Claude-3.7-Sonnet; all of which are available in SissyGPT!
Roleplay Models
Many of the top labs attempt to create something of a "god model" that knows everything and can do everything. Other times, companies may take more experimental approaches with their open source models. Models such as DeepSeek-V3, DeepSeek-R1, Mistral-Small, Mistral-NeMo, and Hermes-3 are renowned for their particularly great roleplay capabilities, with almost no censorship whatsoever.
Emotional Intelligence
Sometimes it's not just about calculating bleeding-edge math calculations or coding performance. Many have noted that models such as Claude-3.5, Claude-3.7, and Gemma-3 are **incredibly emotionally intelligent**, leading them to approach some goals in different ways than others would. Sissy training is about more than just following a task list, and not all sissy trainers are created equal! It's lots of fun to quickly experiment with each model and find the mistress that suits you.
Context Length
Back in the early days of the SissyGPT beta, the best model in the world was the original GPT-4 (which had a max token limit of 8,192 tokens). This was an extremely limiting bottleneck. The model couldn't keep track of what's happening after just a few chat turns. I'm glad to report that we've come a very long way since then. Most of the models available in SissyGPT today have a context window from 128k-200k tokens. Certain models like Gemini-2.5-Pro and Gemini-2.5-Flash can process up to 1,000,000 tokens at once!
Full List of Supported SissyGPT Models
GPT-4.5 | GPT-4.1-mini | GPT-4o-mini | GPT-4-Turbo | Gemini-2.5-Pro | Gemini-2.5-Flash | Gemma-3-27B | Gemma-3-12B | Qwen3-235B-A22B | Qwen3-32B | Grok-4 | Grok-3 | Grok-3-mini | DeepSeek-V3 | DeepSeek-R1 | DeepSeek-R1-0528 | Mistral-Small-3.1 | Mistral-NeMo | Mistral-Medium-3 | Mistral-Large-2 | Llama-4-Maverick | Llama-4-Scout | Claude-3.7-Sonnet | Claude-3.5-Sonnet | Hermes-3-405B | Hermes-3-70B
Image Generation in SissyGPT
SissyGPT is now capable of producing images directly in the chat! Your trainer may choose to do so at will, if you deserve it...
Playground Capabilities Merged
SissyGPT is now capable of everything that our previous API Playground could do! Including the ability to set a custom system prompt and access the raw, foundational models behind the product, with extremely minimal censored outputs.
I'm also pleased to say you can finally export and import your chat conversations and saved prompts in .json format! Make backups of your chats/prompts, duplicate them across devices, jump into an existing chat someone shared with you, the control is all yours!
LLM Lab
I'm excited to introduce the LLM Lab! Previously, the AI Workshop was home to several individual web apps that each provided access to a notable open source model, with all the knobs and sliders you need to create a highly personalized AI personality, including System Prompt, Temperature, Top-P, Frequency Penalty, and Seed. The LLM Lab combines all of them into one app, with the ability to choose from over 6,000 models!
HornyHub
You can now choose any model available on Hugging Face (including custom LoRAs) inside HornyHub. This expands the image generation roster to over 38,000 models! The Featured Models menu has been updated to showcase a wider range of capabilities, including dozens of FLUX LoRAs, the full HiDream-I1 and Stable-Diffusion-3.5 families of models, and many custom fine-tunes. Here are some examples~
Groq Playground
The Groq Playground has been updated with the latest models, at the fastest speeds possible.
New models available in the Groq Playground:
llama-4-maverick-17b-128e-instruct
llama-4-scout-17b-16e-instruct
deepseek-r1-distill-llama-70b
We're happy to welcome the Llama-4 family of models into the Playground, along with a distilled DeepSeek model. This model is significantly smaller than DeepSeek-R1, but was fine-tuned on R1's outputs. Combined with Groq's unique LPUs, it provides world-leading intelligence before other models have even finished reasoning!
Notable Mentions and Bug Fixes
There have been numerous stability improvements and bug fixes across all Alyxsissy's apps, including Virtual Try-On and the TTS Console.
The API Playground is being deprecated as SissyGPT is now the home for this functionality. The new unified structure of SissyGPT and the LLM Lab are much more maintainable, and we intend to keep the featured models in each app updated as notable models are released.
Looking Forward
Everything mentioned above is available TODAY! The next step for SissyGPT is the ability to upload images, it's long overdue. As we're no longer bound by OpenAI's API, we have a bit more freedom.
We are also currently cooking up a more elegant solution to Virtual Try-On and AI-based image editing in general. More to say on these very soon, sissies!