Montalte/qwen3_4b_nh025_thinkcode_a_planb_trainable
TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:4BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Sep 6, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold
Montalte/qwen3_4b_nh025_thinkcode_a_planb_trainable is a 4 billion parameter model based on the Qwen3 architecture, developed by Montalte. This model is a 'Plan B Localize-and-Stitch' merged checkpoint, indicating it's a result of specific local evaluation runs. Its primary characteristic is being a trainable checkpoint, suggesting its utility for further fine-tuning or specialized applications.
Loading preview...
Model Overview
Montalte/qwen3_4b_nh025_thinkcode_a_planb_trainable is a 4 billion parameter model built upon the Qwen3 architecture. This particular version represents a "Plan B Localize-and-Stitch" merged checkpoint, derived from local evaluation runs conducted by Montalte.
Key Characteristics
- Architecture: Based on the Qwen3 model family.
- Parameter Count: Features 4 billion parameters.
- Development Status: Identified as a 'merged checkpoint' from specific local evaluation runs.
- Trainability: The
_trainablesuffix in its name indicates that this model is designed to be further trained or fine-tuned for specific tasks or datasets.
Potential Use Cases
This model is particularly suitable for developers and researchers who:
- Require a Qwen3-based model for further fine-tuning on custom datasets.
- Are interested in experimenting with a checkpoint resulting from a 'Localize-and-Stitch' merging strategy.
- Need a trainable base for specialized applications where a 4B parameter model is appropriate.