google/gemma-3-1b-it
Run locally on Apple devices with Mirai
- Type
- local
- From
- Quantization
- No
- Parameters
- 1B
- Size
- 1.9 GB
- Source
Gemma 3 1B IT is the instruction-tuned variant of Google's smallest Gemma 3 model, part of a family of lightweight, open-weight models built on the same research and technology behind Google's Gemini. Despite its compact 1-billion parameter size, it brings many of the architectural advances of the broader Gemma 3 lineup to resource-constrained environments.
Key Capabilities
This model handles text input and generates text output, making it well-suited for a range of generative tasks including question answering, summarization, and reasoning. As an instruction-tuned model, it has been fine-tuned to follow natural language instructions, making it more immediately useful in conversational and task-oriented workflows compared to its pre-trained counterpart.
Gemma 3 1B IT supports a 128K token context window, enabling it to process lengthy documents or extended conversations in a single pass. It also offers multilingual support across more than 140 languages, broadening its applicability for global and cross-lingual use cases.
Why It Stands Out
The 1B parameter size is the entry point of the Gemma 3 family, designed specifically for deployment where compute and memory are limited — laptops, desktops, edge devices, or modest cloud infrastructure. Despite its small footprint, it benefits from the same architectural improvements shared across Gemma 3, including the large context window and broad language coverage that larger variants offer.
Use Cases
- Lightweight conversational assistants
- On-device text generation and summarization
- Multilingual question answering
- Prototyping and experimentation with limited hardware
- Edge deployment scenarios where latency and resource efficiency matter
Gemma 3 1B IT represents Google's commitment to democratizing access to capable AI models, offering a practical balance between performance and efficiency for developers and researchers working within real-world constraints.
spm https://github.com/trymirai/uzu.git
| 1 | import Uzu |
| 2 | |
| 3 | public func runChat() async throws { |
| 4 | let engineConfig = EngineConfig.create() |
| 5 | let engine = try await Engine.create(config: engineConfig) |
| 6 | |
| 7 | guard let model = try await engine.model(identifier: "google/gemma-3-1b-it") else { |
| 8 | return |
| 9 | } |
| 10 | for try await update in try await engine.download(model: model).iterator() { |
| 11 | print("Download progress: \(update.progress())") |
| 12 | } |
| 13 | |
| 14 | let messages = [ |
| 15 | ChatMessage.system().withText(text: "You are a helpful assistant"), |
| 16 | ChatMessage.user().withText(text: "Tell me a short, funny story about a robot") |
| 17 | ] |
| 18 | let session = try await engine.chat(model: model, config: .create()) |
| 19 | let stream = await session.replyWithStream(input: messages, config: .create()) |
| 20 | var message: ChatMessage? = nil |
| 21 | for try await update in stream.iterator() { |
| 22 | switch update { |
| 23 | case .replies(let replies): |
| 24 | message = replies.last?.message |
| 25 | case .error(let error): |
| 26 | print("Error: \(error)") |
| 27 | } |
| 28 | } |
| 29 | print("Text: \(message?.text() ?? "empty")") |
| 30 | } |