Gumloop Chew
AI model · Language model (LLM) · Text → Text by Gumloop
The intelligent model router for AI agents.
It allows developers to optimize for performance and cost by switching models based on task complexity.
Model specs
The full spec card (context, pricing per token, modalities and more) on AntSpace. Sign up free to see the full spec card.
- Languages
- English
Category: AI Infrastructure
- Modalities
- Text to text
- Sizes and variants
- Gumloop Chew (gumloop-chew)
- License
- Proprietary
- Access
- Gumloop Agent Builder, API, CLI
- Model Identifier
- gumloop-chew
- Model Type
- Router
- Dynamic per-message model routing
- Optimizes for task-specific performance
- Accessible via CLI, API, and Agent Builder
- Identified by model name 'gumloop-chew'
Overview
Gumloop Chew is a proprietary model router developed by Gumloop to optimize AI agent performance and efficiency. It serves as an intelligent selector that picks the most appropriate underlying AI model for a specific message or task rather than relying on a single fixed model. This allows agents to leverage different models dynamically based on the complexity of the request. It is accessible through the Gumloop agent builder, API, and CLI, where it is identified by the model name 'gumloop-chew'. The router can also be used independently via the command line to discover which model is best suited for a particular task.
Model Selection and Routing
- Dynamic Model Selection: The router analyzes the user's task and standing instructions to select the most suitable model for the specific request.
- Per-Message Routing: The router picks a model per message rather than using a single fixed model for the entire session.
- Task-Based Routing CLI: Users can ask the router to choose a model for a task via the terminal without running the task itself.
Integration and Access
- Cross-Platform Access: The router is accessible through the Gumloop agent builder, API, and CLI using the model name gumloop-chew.
- Agent Integration: It can be used as the primary model for an agent by selecting it in the agent configuration.
- Independent Model Calls: The router can be called independently for model selection using the gumloop models route command.
Use cases
- Developers use Gumloop Chew to automatically switch between frontier and faster models based on task complexity.
- Businesses use the router as the primary engine for agents to ensure the most capable model handles reasoning-heavy steps.
Specifications
- Modalities
- Text to text
- Sizes and variants
- Gumloop Chew (gumloop-chew)
- License
- Proprietary
- Access
- Gumloop Agent Builder, API, CLI
- Model Identifier
- gumloop-chew
- Model Type
- Router
- Routing Granularity
- Per-message
How it works
- Configure Agent Model: Select 'gumloop-chew' as the model for your agent in the Gumloop builder or via the CLI.
- Analyze Task Context: When the agent runs, Gumloop Chew analyzes the incoming message and the agent's instructions.
- Select Model per Message
FAQ
What is Gumloop Chew?
Gumloop Chew is a model router that picks a model per message rather than using a single fixed model for the entire session.
How do I use Gumloop Chew in my agents?
You can use it by selecting 'gumloop-chew' as the model name when creating or updating an agent in the builder, API, or CLI.
Does using Gumloop Chew cost credits?
Yes, routing can consume credits as part of the Gumloop billing system.
Can I test the router from the command line?
You can use the 'gumloop models route' command followed by your task description to see which model the router recommends.
Added by AntSpace from public sources (docs.gumloop.com).