Supported Models
BrowserAI comes with a carefully curated selection of models optimized for browser-based AI. Here’s what you can use today.
Text Generation Models
| Model | Description | Best For | Memory Required |
|---|---|---|---|
| smollm2-135m-instruct | 135M parameters | Quick tasks, Getting started | 78MB |
| smollm2-360m-instruct | 360M parameters | Balanced performance | 210MB |
| smollm2-1.7b-instruct | 1.7B parameters | Better comprehension | 1GB |
| llama-3.2-1b-instruct | 1B parameters | General purpose | 712MB |
| hermes-llama-3.2-3b | 3B parameters | General purpose | 1.76GB |
| llama-3.2-3b-instruct | 3B parameters | General purpose | 1.76GB |
| qwen2.5-0.5b-instruct | 500M parameters | Efficient inference | 278MB |
| qwen2.5-1.5b-instruct | 1.5B parameters | Better reasoning | 868MB |
| qwen2.5-3b-instruct | 3B parameters | Better reasoning | 1.7GB |
| gemma-2b-it | 2B parameters | Advanced tasks | 1.44GB |
| phi-3.5-mini-instruct | Compact model | Quick responses | 2.1GB |
| tinyllama-1.1b-chat | 1.1B parameters | Chat interactions | 800MB |
| deepseek-r1-distill-qwen-7b | 7B parameters | General purpose | 4.18GB |
| deepseek-r1-distill-llama-8b | 8B parameters | General purpose | 4.41GB |
Speech Models
| Model | Type | Best For | Memory Required |
|---|---|---|---|
| whisper-tiny-en | Speech Recognition | English transcription | 256MB |
| whisper-base-all | Speech Recognition | Multi-language transcription | 512MB |
| whisper-small-all | Speech Recognition | Improved accuracy | 1GB |
| kokoro-tts | Text-to-Speech | Natural speech synthesis | 512MB |
Embedding Models
| Model | Size | Best For | Memory Required |
|---|---|---|---|
| snowflake-arctic-embed-s | Small (66MB) | Efficient embeddings | 128MB |
| snowflake-arctic-embed-m | Medium (218MB) | Better quality embeddings | 256MB |
Multimodal Models (Coming Soon)
| Model | Description | Best For | Memory Required |
|---|---|---|---|
| janus-1.3 (Coming Soon)b | 1.3B parameters | Vision-language tasks | 1GB |
Coming Soon
We’re constantly working on adding new models and improving existing ones. Stay tuned for:
- More specialized task models
- Improved multilingual support
- Smaller and faster variants