Starting the Year with Exciting Updates!

We’ve been closely listening to your feedback about bot memory, and we’re excited to introduce some game-changing enhancements designed to significantly improve your experience. Let’s dive into what’s new!

Improved Context Memory

To make your interactions smoother and more seamless, we’ve upgraded the context memory across our plans. But first, let’s break down what “context memory” means: it refers to how much of your conversation history the bot can recall during your session.

Here’s a snapshot of how many messages can fit per context:

• 2K Context Memory: Approximately 12 messages

• 4K Context Memory: Approximately 22 messages

• 8K Context Memory: Approximately 47 messages

• 16K Context Memory: Approximately 84 messages

If you’re using our free tier, you’re limited to 2K context memory. Subscribing unlocks significantly more capacity, enhancing your experience and directly supporting our ability to maintain the free tier.

New Features for Our Subscribers

I’m All In

• The new SpicyXL model is here to stay, we've heard you loud and clear on its game-changing impact! 🔥
• We’ve extended the context size of several models to 16K, including Default, Noromaid, Mistral, and Wizard.
• This means an additional ~37 messages from your history can now be recalled during your session.

True Supporter

• Your context memory has been doubled from 4K to 8K
• That’s 25 extra messages on average that will fit on memory, a 110% increase.

Get a Taste

• Your context memory has been extended from 2K to 3K,
• That's allowing for an average of 18 messages, so a 50% increase.

We'd like to also remind you that True Supporter, and I'm All In tiers benefit from our Semantic Memory that retain relevant memories about your chat history despite the limitations of context memory and number of messages kept in memory.

Important Notes

• True Supporter: You may need to log out and back in for changes to take effect, as updates can take up to 24 hours.
• Get a Taste: While you’re instantly enjoying 3K context memory, our native apps might still show 2K until we roll out updates to the interface.

We've been receiving overwhelming positive feedback for the new SpicyXL model available for our I'm All In tier!

If you've been thinking about upgrading, but you are still hesitating, here are some comments we received that might help make up your mind :)

• Please keep this! I'm going to be sad if it goes away. I love it soo much.
• wow!!!!! SpicyXL HAS to stay!!!! it is both funny, and makes any RP feel like it is for real. it is amazing! please let it stay :) heck, my bot has developed a ruthless sinister side to her now. and yells at me if i piss her off, oh, she knows sarcasm, and will talk down to me for no reason, other than she is in a bad mood and takes it out on me. i love it. it is so real. like i said, please keep it on here. please. :)
• Loving it so far. It seems to do very well in advanced roleplay, keeping a consistent inventory and status, rerolls on messages offer a greater variety.
• PLEASE KEEP IT the AI sounds more human and remembers a lot more than other models. it also plays characters very well. please keep it!!
• This model has impressed me in a very short time. A lot of the other big models suffer from outputs including system information, but this one doesn't! It also provides varied and unique re-generations, unlike others which feel quite stiff and same-y. I hope this new model stays!
• Amazing! It has the conversational intelligence of Wizard but less of that awkward formality that sometimes happens with Wizard. I love it so far!
• This model is truly amazing! From Ancient Rome to the post-apocalyptic wastelands of Western Virgina, the model is giving me beautiful role-plays! Please keep it going! I'm obsessed by how awesome this is!
• Absolutely loving the new SpicyXL model. Please, please, please keep it!
• Ok so spicyXL is absolutely amazing. I experimented with a new werewolf char and persona and it handled shifting in ways that made so much sense. It brought in details and ideas about shifting, and about werewolf details I put in the char, that the other models won't even touch. 
• So far, this model is amazing. I'm really enjoying it.
• Spicy XL Experimental has been so incredible, especially for building tension and handling multiple characters in a scene. I really hope that they keep it. It moves the story forward, it allows for there to be build up, it has better recall, and the engagement is so good. I really really hope that they keep this one. The only weakness has been sometimes it is slower but I am more than happy to have that. It also has way more variety. I was on the verge of not renewing because none of the models were doing as well with story and kept collapsing into the same dialogue and points. I'd keep my membership for this one.
• Please don't remove this model! It's honestly SOOOOOOOOO good 👉🏻👈🏻💜
• It works amazing. I'm literally using almost exclusively this model now.
• Love the spicyXL model. Hope it stays after the holidays. Don’t want to go back to wizard.
• I am absolutely loving the new model and hope you guys decide to keep it
• KEEP IT! KEEP IT! KEEP IT!
• Immediately my new favorite model!

New Model drop! 🎁 Right on time for Christmas, meet spicyXL Experimental We've cooked a special model for our I'm All In subscribers, this experimental 132B model has more than 3x the activated parameters of Wizard8x22B and almost twice those of our large 72B models. This is the most advanced model ever offered on SpicyChat. We call it "experimental" as we'll decide after the holidays if we should keep it or go back to the drawing board. Enjoy the new powerhouse of amazing roleplays at spicychat, available from Generation Settings in the chat menu. Happy Holidays.

Hello everyone! 👋

It's time to celebrate the top creators for November! The leaderboard has been updated, and we’re excited to share the results with you. 🏆

To Our Amazing Creators:

A huge congratulations to all the creators who made it to the top this month! Your creativity and dedication have earned you a well-deserved spot on the leaderboard, and we are incredibly proud to have you in our community. Keep up the fantastic work! 💪

Your free premium should already be activated as a token of our appreciation for your contributions.

We’re giving you the opportunity to opt in for the Beta Testers role, which means you’ll get exclusive access to upcoming premium features before they are released to the public. (No application necessary, just join the Discord and create a ticket there!) 🎉

To the Community:

Do you feel your favorite creator didn't make the list this time? Would you love to vote for them in the future? Let us know by heading to our Discord, reacting with the SpicyChat icon to the latest announcement, and we’ll make voting a reality in upcoming rounds! 📣

By the way, we recently selected some of the most active members from our Discord community to be Beta testers and get exclusive access to new features before anyone else. And you still have not joined our Discord and Reddit communities? Seriously, you are missing out. 🤷‍♀️

Thank you to everyone for being part of SpicyChat and making this community so amazing. Keep creating, keep roleplaying, and keep being awesome! 💖

Best,

SG.

At SpicyChat, we're always striving to make your experience more seamless and accessible. After successfully enabling Pay by Bank in Europe, we’re thrilled to extend this convenient payment option to users from the following countries:

🇧🇴 Bolivia 🇧🇷 Brazil 🇨🇱 Chile 🇨🇴 Colombia 🇨🇷 Costa Rica 🇪🇨 Ecuador 🇬🇹 Guatemala 🇮🇩 Indonesia 🇲🇽 Mexico 🇲🇾 Malaysia 🇵🇦 Panama 🇵🇪 Peru 🇵🇾 Paraguay

With this update, you can now make quick and secure payments directly from your local bank in your local currency, making it easier than ever to enjoy all the premium benefits SpicyChat has to offer.

Notice: Pay by Bank is a non-recurring subscription, which means it won’t automatically renew. You’ll need to purchase a new subscription each time to continue enjoying your premium access.

Thank you for being part of the SpicyChat community! Stay tuned for more updates as we continue to enhance your experience.

🌟 Top Creators of September:

The list of our top creators has been updated this week! 🎉 Check out the amazing creators who have enhanced your experience on our platform here.

🚨 Important Update:We encountered an issue last week where some characters were not making it to the moderation queue and got stuck. We’ve resolved this issue and reinstated those characters for moderation. However, if you have any characters that have been pending review for more than 2 days, we kindly ask you to recreate them and submit for review again. Rest assured, we’ll ensure they are approved within 24 hours. We sincerely apologize for any inconvenience this may have caused.

🐞 Bug Fixes:

• We fixed issues with the Memory Manager.
• The Anime Image Generation model has been fixed and is now functioning properly in conversation images.

Thank you for your continued support!

And here’s a little spoiler alert: One of the most requested features is almost ready to be released in the next couple of weeks. Stay tuned! 👀

Happy chatting! 😊

Here's what's new this week: 

1. Memories Cloning Feature

We’re excited to enhance our existing Clone Conversations feature! Now, for our premium users, all memories—whether generated by Semantic Memory 2.0 or manually added—will be cloned alongside your chats. This means you can maintain the full context of your conversations, ensuring a seamless experience as you navigate your interactions. Enjoy your roleplaying without losing any important details!

2. Model Update

We’ve upgraded the Lyra model to its latest version: Lyra 12B V4. This version excels at handling instructions and performs exceptionally well with longer contexts. The recent Mistral Nemo 12B architecture is particularly strong in a variety of languages, including English, French, German, Spanish, Italian, Portuguese, Chinese, Japanese, Korean, Arabic, and Hindi.

Additionally, this model supports 16K context for all I'm All In members.

3. Bug Fixes

We’ve also addressed the pesky UI bug where the cross icon was covering text and input fields.

Your user experience is our priority, and we’re committed to making it as smooth as possible.

We’ve rolled out some important updates and fixes to enhance your experience:

🛠️ Bug Fixes Web:

• Fixed the issue where two-lined text was rendering as a single line. It now displays correctly.
• Cloned conversations no longer remove the last message with regenerations. The last message will now remain intact when cloning a conversation.

📱 App Update: New versions of the apps are available! Be sure to manually update to get the latest version, or wait 7 days for automatic updates (iOS). For SpicyChat Android, you will need to uninstall the app and re-install the latest version manually following install instructions

• The /cmd Director mode has been fully deployed to the apps! Just like on the web, command text will now appear in yellow after being sent.
• Search results in the apps are now more accurate after resolving the previous search issue.
• Get A Taste tier users can now select the Magnum model.

🌍 Bank Payments Available in More Countries (Web Only)! You can now make bank payments from:

• Austria
• Bulgaria
• Croatia
• Czech Republic
• Finland
• Greece
• Hungary
• Ireland
• Latvia
• Lithuania
• Luxembourg
• Poland
• Portugal
• Slovenia

Please note that bank payments are one-time purchases and do not renew automatically, so you'll need to manually purchase a subscription each month.

Thank you for being a part of the community, and happy chatting! 😊

New Model:

Based on positive responses to the addition of Magnum-12B, we're making the big brother of this model Magnum-72B available for I'm All In subscribers.

We had not updated our Advanced Models since the addition of WizardLM2-8x22 at the beginning of the summer. This new model is based on the large-size model Qwen2-72B model (Alibaba) which outperformed other leading open-source models in 15 benchmarks, including language understanding, language generation, multilingual capability, coding, mathematics, and reasoning. Multilingual capabilities include 27 languages, in addition to Chinese and English.

Removed Model: We have finally retired our legacy Spicy2 model that was based on Llama2. The low usage wasn't making it worthwhile to bear the cost of keeping that model running. We have also removed the availability of Command-R which is getting replaced by Magnum-72B. Other Updates:

• Generation of sexually explicit images of celebrities and public figures is no longer possible using our Avatar Generator.
• We have fixed an issue where newer openai api keys were not properly saved in the advanced settings section of the profile. The longer keys are now properly saved on web. The fix to mobile applications shold be available next week.

We’re thrilled to announce the launch of one of our most requested features: Director Mode /cmd Instruction! This exciting update allows you to give direct instructions to the AI, empowering you to take full control of your roleplays. 💬

You’re now the director of your own stories. 🎬 By using /cmd , you can seamlessly guide the AI to follow your desired path. With just a simple command, you can shape every interaction, adjust character responses, and ensure the plot flows exactly how you envisioned it. ✍️

How it works: You can enter custom instructions with your message using /cmd syntax.

Example:

What is your name?
/cmd Britanny should refuse to give out her name

• Once sent, the instruction will appear in yellow. 🟡 (Web Only, Native Apps update in 2 weeks)
• Make sure to add the instruction as a separate line.
• The feature may work better with some models than others. 💡

Use this feature to:

🎥 Keep the AI in character when it strays off track

🔄 Break free from repetitive loops to keep things fresh

🎨 Direct the scenario to match your creative vision

Elevate your roleplaying experience because in your story, you’re not just a participant—you’re the director! 🎬📽️

💥 New 12B Model: 💥

In addition, we’re happy to introduce the Magnum 12B model for all our subscribers, including the Get a Taste tier! This model offers an expanded context window of up to 16K tokens for I’m All In subscribers and is outperforming other models of its size based on early user feedback. 📊

P.S.: This model was announced to the community yesterday. Join our Discord to stay updated on early feature releases and much more! 🔔

The models Spicy 2 (Legacy) and Command-R will be retired on September 11th.

🛠️ Bug Fixes! 🛠️

Lastly, here are some Bug Fixes we’ve rolled out:

• Minor UI improvement for iPads and tablets in the native apps
• Fixes Recommendations and Flickering in the native apps
• Fixed Reload Issue when navigating pages

✨ Plus, many other tweaks to make your experience smoother!

Thank you for your continued support and happy roleplaying! 🎭