Mindcraft-CE/Andy-4.2-Air
Mindcraft-CE/Andy-4.2-Air is a 4.5 billion parameter language model from Mindcraft CE, built on the Qwen3.5 architecture with Gated Deltanet attention. It is specifically optimized for playing Minecraft, demonstrating the ability to achieve a full set of diamond armor with zero human interaction. This model features vision capabilities for deep game state comprehension and supports a context length of up to 1 million tokens, making it highly efficient for its specialized task.
Loading preview...
Overview
Mindcraft-CE/Andy-4.2-Air is a 4.5 billion parameter model developed by Mindcraft CE, utilizing a new Qwen3.5 architecture with Gated Deltanet attention. This design makes it more efficient and intelligent than previous iterations, capable of running on a single RTX 3090 with 256k tokens of context at 8-bit quantization.
Key Capabilities
- Minecraft Automation: The first local model capable of acquiring a full set of diamond armor in Minecraft without human intervention.
- Enhanced Vision: Features stronger multimodal capabilities for deeper comprehension of the game state.
- Efficiency: Runs efficiently on consumer-grade hardware (e.g., single RTX 3090) due to its optimized architecture.
- Large Context Window: Supports a context length of up to 1 million tokens, with stable performance up to 120 messages.
Limitations
- Building Tasks: Not advised for building tasks within Minecraft, as it tends to produce extensive output without executing actions.
- Code Generation: Not suitable as a general-purpose code model.
Recommended Use Cases
- Automated Minecraft Gameplay: Ideal for scenarios requiring autonomous interaction and progression within Minecraft.
- Game State Analysis: Useful for applications needing deep understanding and response to complex game environments.