Maid is a cross-platform Flutter app for interfacing with GGUF / llama.cpp models locally, and with Ollama models remotely. The goal of Maid is to create a platform for AI that can be used freely on any device.
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
Windows, Linux, Android. Releases page. MacOS and IOS Releases not available at this time.
| Parameter | Local | Ollama | OpenAI | MistralAI |
| model | Y | Y | Y | Y |
| seed | Y | Y | Y | Y |
| temperature | Y | Y | Y | Y |
| responseFormat | N | Y | Y | N |
| numKeep | N | Y | N | N |
| numPredict | N | Y | N | N |
| topK | Y | Y | N | N |
| topP | Y | Y | Y | Y |
| tfsZ | Y | Y | N | N |
| typicalP | Y | Y | N | N |
| repeatLastN | Y | Y | N | N |
| repeatPenalty | Y | Y | N | N |
| presencePenalty | Y | Y | Y | N |
| frequencyPenalty | Y | Y | Y | N |
| stop | N | Y | Y | N |
| mirostat | Y | Y | N | N |
| mirostatTau | Y | Y | N | N |
| mirostatEta | Y | Y | N | N |
| penalizeNewline | Y | Y | N | N |
| numa | N | Y | N | N |
| numCtx | Y | Y | N | N |
| numBatch | Y | Y | N | N |
| numGqa | N | Y | N | N |
| numGpu | N | Y | N | N |
| mainGpu | Y | Y | N | N |
| lowVram | N | Y | N | N |
| f16KV | N | Y | N | N |
| logitsAll | Y | Y | N | N |
| vocabOnly | Y | Y | N | N |
| useMmap | Y | Y | N | N |
| useMlock | Y | Y | N | N |
| embeddingOnly | Y | Y | N | N |
| ropeFrequencyBase | Y | Y | N | N |
| ropeFrequencyScale | Y | Y | N | N |
| numThread | Y | Y | N | N |
| maxTokens | N | N | Y | Y |
| n | N | N | Y | N |
| logitBias | Y | N | Y | N |
| functions | N | N | Y | N |
| functionCall | N | N | Y | N |
| user | N | N | Y | N |
| safeMode | N | N | N | Y |
- Write code comments
- Documentation
- Testing and Building on MacOS and IOS
- Spreading the word
Android version tested on a Oneplus 10 Pro 11gb phone. Also tested on Fedora Linux, Windows 11. Tested with calypso 3b, orcamini 3b, llama 2 7B-Chat, llama 7B and many more.
Please note that the llama.cpp models are owned and officially distributed by Meta. This app only serves as an environment for the model's capabilities and functionality. The developers of this app do not provide the LLaMA models and are not responsible for any issues related to their usage.