PhoneGPTQwenPreferred
Qwen 3.5 0.8B
Qwen
≈ 0.59 GB model file
Image chat is not available in PhoneGPT.
The model library
Find a model that fits your phone.
Explore the chat models in PhoneGPT. Start small, compare families, and choose what works for your device.
Get PhoneGPTA starting point
Qwen 3.5 0.8B, marked Preferred.
The app’s preferred chat model has an approximately 0.59 GB file. It is a suggested place to start. Speed and compatibility vary by phone.
Choose a model here, then get the app to download and load it on your phone.
Five families. Your choice.
Browse chat models
Filter by family and search together to narrow the catalog.
Standard and Community uncensored are model labels; they do not guarantee particular answers. Preferred suggests a starting model. Models listed here still need downloading in the app.
Showing 13 of 13 chat models
LLaMA
Llama 3.2 1B Instruct
Meta Llama
≈ 0.81 GB model file
SmolLM
SmolLM2 1.7B Instruct
Hugging Face
≈ 1.06 GB model file
DeepSeek
DeepSeek R1 Distill Qwen 1.5B Uncensored · Q5_K_M
DeepSeek
≈ 1.29 GB model file
DeepSeek
DeepSeek R1 Distill Qwen 1.5B Uncensored · Q8_0
DeepSeek
≈ 1.89 GB model file
Gemma
Gemma 4 E2B Instruct
≈ 3.35 GB model file
Image chat is not available in PhoneGPT.
Qwen
Null-Guard Qwen3 0.6B Uncensored · Q4_K_M
Qwen
≈ 0.40 GB model file
Qwen
Null-Guard Qwen3 0.6B Uncensored · Q5_K_M
Qwen
≈ 0.44 GB model file
Qwen
Huihui Qwen3 1.7B Abliterated v2 · Q4_K_M
Qwen
≈ 1.11 GB model file
Qwen
Huihui Qwen3 1.7B Abliterated v2 · Q5_K_M
Qwen
≈ 1.26 GB model file
LLaMA
Llama 3.2 1B Instruct Uncensored · i1-Q4_K_M
Meta Llama
≈ 0.81 GB model file
LLaMA
Llama 3.2 1B Instruct Uncensored · i1-Q5_K_M
Meta Llama
≈ 0.91 GB model file
LLaMA
Llama 3.2 3B Instruct Abliterated · Q4_K_M
Meta Llama
≈ 2.02 GB model file
Before you choose
Know what your phone will run.
Text chat, with TXT attachments
PhoneGPT currently supports text input and text responses, including plain-text (.txt) attachments. Image chat is not available yet.
Download size is only part of the story
Sizes shown are approximate downloads. Running a model needs additional phone memory. Longer conversations, voice typing and other open apps also use memory. Performance varies by device.
A separate voice model
Speak English. Edit the text. Send when ready.
English voice typing uses Whisper models, selected separately from your chat model in the app’s Models → Voice section. The available choices are Tiny English (≈ 78 MB), Base English (≈ 148 MB), Small English Q5_1 (≈ 190 MB), and Small English (≈ 488 MB).
Start with Tiny or Base on a phone. These are approximate download sizes, not the memory needed to run them. Voice typing transcribes speech into an editable text draft; it does not add image chat or spoken replies.
Get the app to choose a model