Disclosure: we make Pinku, the app this page is about. The figures come from Pinku's public model catalog (updated 31 October 2026) and the model's Hugging Face card.
| Made by | NVIDIA |
|---|---|
| Parameters | 8 billion (4 billion effective) |
| Download size in Pinku | 7.5 GB |
| Memory needed to be listed | 16 GB |
| Shows its reasoning | No |
| Licence | OpenMDW 1.1 |
| Model card | nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-BF16 |
| Build Pinku downloads | mlx-community/NVIDIA-Nemotron-3.5-Bolt-E4B-OptiQ-4bit |
| 5-question test | 5 of 5 |
| In Pinku | Pro |
What it is good at
- Chatting and writing
- Writing and summarizing
As described in Pinku's catalog.
Pinku's developers ask new candidate models five factual questions on a Mac (capital of France, days in a leap year, author of Don Quixote, chemical symbol for gold, largest planet). It is five questions, not a benchmark. 5 of 5.
The first answer after loading is slow to start: on an M3 Mac with 16 GB we measured from under 30 seconds to about a minute. Pinku warns about the slow start in the model list.
Devices that run it
Pinku offers this model only on a Mac. It is not listed on iPhone or iPad.
Mac: Any Mac with Apple silicon and 16 GB of memory or more, on macOS 14 or later.
Which model fits my device? (tool)
How to try it in Pinku
- Install Pinku on your Mac from the App Store.
- Open the model list; Pinku only shows models your Mac has memory for.
- Choose Nemotron 3.5 Bolt E4B. It needs Pinku Pro (yearly subscription or one-time purchase); chatting with Llama 3.2 1B is free.
- Download it once. After that it works offline.
Other models in Pinku
- Qwen3 0.6B · 0.4 GB
- LFM2.5 1.2B · 0.7 GB
- Llama 3.2 1B · 0.7 GB
- Gemma 3 1B · 0.8 GB
- Qwen3 1.7B · 1 GB
- Qwen3.5 2B · 1.7 GB
- Llama 3.2 3B · 1.8 GB
- SmolLM3 3B · 1.8 GB
- Qwen3 4B · 2.3 GB
- Gemma 3n E2B · 2.6 GB
- Gemma 4 E2B · 3.6 GB
- Qwen3 8B · 4.7 GB
- Astrea R8 9B · 8 GB
Questions people ask
Can my iPhone run Nemotron 3.5 Bolt E4B?
Only on a Mac: Pinku lists it on Macs with Apple silicon and at least 16 GB of memory, not on iPhone or iPad.
Is Nemotron 3.5 Bolt E4B free in Pinku?
It is one of the models that come with Pinku Pro, a yearly subscription or a one-time purchase. Chatting with Llama 3.2 1B is free.
Does Nemotron 3.5 Bolt E4B work offline?
Yes. In Pinku you download it once (7.5 GB) and every answer is then generated on the device, with no connection.