Loke-60000/gl-agent-1-27b

VISIONPricing:Input $1.6 / Cached $0.15 / Output $12Concurrent Unit Cost:2Model Size:27BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 21, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Loke-60000/gl-agent-1-27b is a 27 billion parameter language model developed by Loke-60000, specifically fine-tuned for long-horizon agentic tasks. It features a native context length of 262,144 tokens, extensible to 1,000,000 tokens with YaRN, and incorporates multi-token prediction (MTP) for speculative decoding. This model excels at tool calling, emitting XML-formatted tool calls, and includes an integrated 'thinking' capability to enhance reasoning.

Loading preview...

Model Overview

Loke-60000/gl-agent-1-27b is a 27 billion parameter model engineered for long-horizon agentic tasks. It is designed to facilitate complex, multi-step operations by integrating advanced capabilities for tool use and internal reasoning.

Key Capabilities

  • Extended Context Window: Features a native context length of 262,144 tokens, which can be expanded up to 1,000,000 tokens using YaRN, enabling the processing of very long inputs and task histories.
  • Robust Tool Calling: Supports structured tool calls, emitting them as XML. This allows for seamless integration with external functions, where the model can dynamically invoke tools like run_bash based on user prompts.
  • Integrated Multi-Token Prediction (MTP): Includes a built-in MTP head for speculative decoding, which enhances inference speed without requiring a separate draft model. This feature can significantly improve wall-clock performance, with a measured mean acceptance length of approximately 1.7 for a single speculative token.
  • Configurable Reasoning ('Thinking'): The model incorporates an 'on by default' thinking mechanism, which can be controlled per request. This allows users to enable or disable internal reasoning processes and adjust the 'reasoning_effort' to suit specific task requirements.

Use Cases

This model is particularly well-suited for applications requiring:

  • Automated Agentic Workflows: Ideal for building AI agents that need to perform sequences of actions, interact with external systems via tools, and maintain context over extended periods.
  • Complex Problem Solving: Its long context and reasoning capabilities make it effective for tasks that involve analyzing large amounts of information and planning multi-step solutions.
  • Interactive Development Environments: Can be used to create intelligent assistants that can debug code, manage configurations, and execute commands within a development environment.

Finetunes
2 models