Introduction to Elephas AI Engine
Elephas supports wide range of AI providers from OpenAI to Gemini.
Elephas, when you subscribe through our website, comes with an inbuilt AI engine, backed by OpenAI APIs. The free account has a limited set of tokens. You can upgrade to the next plan to avail more tokens.
The number of tokens available will vary based on your plan. Check out https://elephas.app/pricing to know the exact number of tokens. All tokens are given per month basis.
We offer the following list of models as part of the subscription,
- gpt-5-mini
- gpt-4.1-mini
- gpt-4o-mini
- gpt-3.5-turbo
- claude-3-haiku
- gemini-2.0-flash-001
- gemma-2-9b-it
- qwen-2.5-72b-instruct
- deepseek-r1-distill-qwen-32b
Using Elephas Inbuilt Offline Models
Elephas also comes with inbuilt offline AI models, available by default as part of your subscription.
These models allow you to run chat and indexing locally on your Mac — ensuring faster performance and complete privacy since no data leaves your machine.
Your subscription includes access to the following offline models:
Chat Models
- Llama 3.2 3B Instruct
- Qwen 2.5 1.5B
- Llama 3.2 1B
- SmolLM 135M
- Qwen 3 4B
- Mistral 7B Instruct v0.3
Indexing Models
- Multilingual E5 Large
- Multilingual E5 Small
💡 You need an Apple Silicon Mac (M1 and later) to use offline models.
For detailed setup, see Using Elephas Inbuilt Offline Models.
Elephas AI model is configured by default.
Elephas does offer the ability to bring your choice of AI providers, and we support popular AI providers.
Here are the current ones,
- LM Studio (Custom AI models)
- Jan.ai (Custom AI Models)
- Groq AI - Fastest inference engine
Here is how the configuration page looks,
Once you configure the necessary AI providers, you can configure the different features with them.
In the above example, I have configured it to use local Llama 3 for chat, rewrite uses gpt-4o from OpenAI, and continue writing uses the Elephas native engine. Thus, you can configure different AI providers based on your use case.