Faraday.dev
Backyard AI is a powerful platform that offers AI-powered chat and interactive story creation tools, capable of working offline with zero configuration. It host...
Last verified:
What is Faraday.dev?
Backyard AI (formerly Faraday.dev) is an AI-powered platform for creating and interacting with immersive text and voice chat characters without filters. It offers both a desktop application for running open-source large language models locally on your hardware and a cloud-based service accessible from anywhere via browser. The platform emphasizes privacy-first principles with local model execution keeping conversations on your device, while also providing cloud options for access to more powerful models.
Key features include Dynamic Voices & Audio for voice conversations with characters, Lorebooks that provide unlimited dynamic memory to enrich the AI's understanding of context history and memories, Author's Note for setting scenes and guiding stories, Grammars to customize response structure and style, and advanced Model Parameters including Min-P, Top-P, Top-K, Temperature, and Repetition Penalty for fine-grained control. The platform supports Group Chats with up to 4 characters at once and Voice Calls for talking to characters.
Backyard AI is designed for users who want AI-powered chat and interactive text adventures including roleplay, storytelling, and character interaction. It appeals to both casual users who want easy access to conversational AI and advanced users who want fine-grained control over model behavior. The platform is suitable for fans of fiction text chats, roleplayers, storytellers, and anyone interested in customizable AI characters with no censorship filters.
Faraday.dev pricing
Pricing model: Freemium
Free tier: No credit card required, 300 messages per week, 7 models, 16k max context size, includes Group Chats, Voice Calls, Lorebooks, and Advanced Model Samplers. Standard plan: $12/month monthly or $9/month annual, unlimited messages, 7 models, 16k max context size, all features included. Pro plan: $35/month monthly or $26/month annual, unlimited messages, 12 models including 70B class models, up to 100k max context size, premium Discord support channel, early access to experimental features, and iOS/Android apps. All plans include a 7-day money-back guarantee.
Faraday.dev pros
- Runs locally on your hardware for privacy and offline access
- No filters or censorship like Character.ai for unfiltered text generations
- 12+ AI models available including 70B class models on Pro plan
- Up to 100k token context size on Pro plan for long conversations
- Dynamic voice generation for audio conversations with characters
- Lorebooks provide unlimited dynamic memory for characters
- Author's Note feature to set scenes and guide story direction
- Grammars enforce custom response styles and structures
- Advanced model samplers including Min-P Top-P Top-K Temperature
- Group chats with up to 4 characters at once
- Voice calls to talk directly to your characters
- 1000+ pre-made characters in the Community Hub
- Zero-configuration setup works immediately out of box
- Supports LLaMA-based models including WizardLM Vicuna Koala
- Web and iOS app available with Android app coming soon
- Private Discord support channel for Pro subscribers
- 7-day money-back guarantee on all subscriptions
- Early access to experimental features for Pro users
Faraday.dev cons
- Free plan limited to 300 messages per week
- Pro plan at $35/month is expensive compared to competitors
- Android app not yet available only iOS and web
- Local usage requires GPU and sufficient hardware resources
- 70B models require significant GPU power to run locally
- Free plan only 16k context vs Pro 100k context
- Only 7 models on Free/Standard vs 12 on Pro
- No native Android app yet announced
- Cloud usage requires internet connection
- Local models must be downloaded and managed manually
Frequently asked questions about Faraday.dev
What is the difference between the plans?
Model selection and context size are the main differences between the Standard and Pro plans. Free has 300 messages per week while Standard and Pro have unlimited messages. Free and Standard have 7 models and 16k context size, while Pro has 12 models including 70B class models and up to 100k context size. If unsure which plan to choose, you can try models on both plans for free with 300 free messages per week.
Are the Pro AI models really that much better?
Yes, the Pro AI models are larger which means they require more GPU power to run but are also more accurate and generate better results. They have a larger context size meaning they can remember more of the conversation which is important especially for long conversations. Pro includes models like Midnight Rose 70B, Llama 3.3 Euryale v2.3 70B, Llama 3.3 Instruct 70B, and Command R+ 104B.
Is Backyard.ai secure and private?
Yes, Backyard.ai is secure and private. They use industry-standard encryption to protect your data. When running locally on your hardware, conversations and generated content remain on your computer rather than being sent to external servers, keeping user data under the user's control.
Is there a filter like Character.ai?
No, they do not have a filter like Character.ai. They do not censor or filter text model generations. However, they have Community Guidelines that everyone must follow when sharing content on the Community Hub.
Do you offer refunds?
Yes. If you don't love Backyard.ai within the first 7 days of your subscription, they will refund you the full amount. This money-back guarantee applies to all subscription plans.
How do I cancel my subscription?
You can cancel your subscription at any time in the Manage Subscription page or by reaching out to them over email at [email protected].
How do I download a character from the Hub?
Clicking on a character takes you to a screen displaying all details and metadata for the character. From here you can download it by clicking the Download button on the top-right of your screen.
How do I post my character to the Hub?
Click on Post to Hub located on the chat page in the right sidebar panel underneath the character image. Fill out the form with a brief description for the character card, add tags to make it easier for people to find, and provide a recommended model which is usually the model you used while making the character. Read the community guidelines to ensure your character meets standards then hit Post Character to upload.
What models are available on each plan?
Free and Standard have 7 models: Mistral-Nemo Instruct 12B, Fimbulvetr v2 11B, Llama 3 Stheno V3.3 8B, Backyard AI Party 12B, Chunky Lemon Cookie 11B, Llama 3 Jamet MK.VBlackroot 8B. Pro has 12 models including all Standard models plus Command R+ 104B, Llama 3.3 Euryale v2.3 70B, Llama 3.3 Instruct 70B, Midnight Rose 70B, Magnum 72B, and Psyonic-Cetacean Ultra 20B.
What are Lorebooks and how do they work?
Lorebooks are a powerful tool to enrich the AI's understanding of relevant context history and memories. Each item in a Lorebook is a keyword with an associated value. When the keyword is used in your chat, the entry is added dynamically to the chat context so that the model understands what you are referring to. You can add multiple associated keywords like home house pad for the same entry.