Soulbits Cloud
A convenience layer on top of our open technology — for when you want the power without the hardware.
Everything Soulbits offers can run on your own machine, for free, forever. Cloud exists for one reason: not everyone has a GPU, and not everyone wants to manage Docker containers. If that's you, we've got you covered — and your data stays encrypted the entire time.
Your Data Stays Encrypted. Always.
When you use Soulbits Cloud, your companion data — conversations, memories, personality — is encrypted on your device before it ever touches our servers. We store opaque ciphertext in encrypted S3 blob storage. Our Session Broker orchestrates containers, but the containers process data that only your client can decrypt. We literally cannot read your data. There is no backdoor.
Cloud Plans
Choose a plan based on how much cloud compute you need. Every plan includes the full feature set — the difference is quotas and model access.
- External Chat Integration
- External Chat Integration
- Internal Translation
- Real Time Image Streaming
- Voice Call Access
- External Chat Integration
- Internal Translation
- Real Time Image Streaming
- Voice Call Access
Available Models
Each model requires a minimum subscription tier. All models can also be run locally for free.
VoiceFixer
voicefixer
Audio restoration and enhancement — repairs degraded or low-quality recordings.
Harrier OSS v1 0.6B
harrier-oss-v1-0-6b
Microsoft's state-of-the-art sub-1B text embedding model (1024-dim).
Qwen3 Embedding 4B
qwen3-embed-4b
Qwen3 embedding model producing 2560-dimensional text vectors — ideal for semantic search and RAG pipelines.
Gemma 4 MeroMero 26B-A4B
gemma4-meromero-26b-a4b
Capable mid-size language model based on Gemma 4 26B (A4B active). Supports text, visual and audio input.
Qwen 3.5 9B
qwen-35-9b
Fast, efficient 9B language model for general chat and visual analysis.
Starfallen Snow Fantasy 24B
starfallen-24b
A very potent model for AI roleplay and character consistency.
Qwen3 Reranker 0.6B
qwen3-rerank-0-6b
Lightweight reranker that scores document relevance against a query — optimized for retrieval-augmented generation.
Whisper Large v3 Turbo
faster-whisper-large-v3-turbo
High-accuracy speech-to-text transcription.
Whisper Tiny
faster-whisper-tiny
Fast and lightweight speech-to-text transcription.
Chatterbox Multilingual
chatterbox_multilingual
Multilingual text-to-speech supporting 23 languages.
Chatterbox TTS
chatterbox
High-quality text-to-speech with voice cloning support.
Chatterbox Turbo TTS
chatterbox_turbo
Optimized Chatterbox variant with lower latency.
HarmonySpeech
harmonyspeech
Fast and lightweight TTS with voice cloning capabilities.
KittenTTS Micro
kitten-tts-micro
Ultra-lightweight text-to-speech, fast.
KittenTTS Mini
kitten-tts-mini
Lightweight text-to-speech, good balance of speed and quality.
KittenTTS Nano
kitten-tts-nano
Smallest KittenTTS variant, very fast.
OpenVoice v1
openvoice_v1
MyShell's reliable text-to-speech model with cross-lingual voice cloning.
OpenVoice v2
openvoice_v2
V2 of MyShell's text-to-speech model with improved quality and voice cloning.
Silero VAD
silero-vad
Ultra-fast AI Voice activity detection — determines if speech is present in audio. Used for Realtime Interactions.
Chatterbox VC
chatterbox_vc
AI Voice conversion — transforms a source voice to match a target voice.
