Nanthasit/sakthai-context-1.5b-merged

Hugging Face
TEXT GENERATIONConcurrent Unit Cost:1Model Size:1.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jul 5, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Warm

Nanthasit/sakthai-context-1.5b-merged is a 1.54 billion parameter Qwen2.5-1.5B-Instruct variant, fine-tuned by Nanthasit for tool-calling and agentic tasks. This model excels at structured tool-calling with XML output, distinguishing between direct answers and tool use, and maintaining multi-turn conversation context. It offers CPU-friendly inference speeds, making it suitable for local-first applications and edge devices.

Loading preview...

SakThai Context 1.5B Merged: Tool-Calling Agent Model

This model, developed by Nanthasit, is a fine-tuned variant of Qwen2.5-1.5B-Instruct, specifically optimized for tool-calling and agentic tasks. It is the most downloaded model in the SakThai family, designed to operate efficiently even on budget-constrained hardware.

Key Capabilities

  • Intelligent Tool Use: Distinguishes when to call tools versus answering directly, supporting multi-turn conversations with tool integration.
  • Structured Output: Emits tool calls in a structured <tool_call> XML format, requiring a <tools> block in the system prompt for reliable function calling.
  • Efficient Performance: Offers CPU-friendly inference speeds (e.g., ~5.23 tokens/sec on CPU for tool-calling), making it suitable for local and edge deployments.
  • Full-Weight Checkpoint: Provided as a merged full-weight checkpoint, eliminating the need for PEFT (Parameter-Efficient Fine-Tuning) during deployment.
  • Self-Verified Tool-Calling: Achieves a self-verified 5/5 tool-calling score on internal benchmarks, demonstrating high accuracy in generating valid tool calls and arguments.

Good For

  • AI Agents: Ideal for building AI agents that require precise tool-calling capabilities.
  • Local-First Applications: Its small size (1.54B parameters) and CPU-friendly inference make it suitable for deployment on edge devices or local machines.
  • Budget-Constrained Projects: Developed with a "zero-cost mindset" on free compute, aligning with projects seeking cost-effective AI solutions.
  • Structured Interactions: Excels in scenarios where structured outputs and clear distinctions between direct responses and tool invocations are critical.