Skip to main content
AI models are evolving rapidly. New models are typically available in Gumloop within one day of their public release, so you’ll have access to the latest capabilities even if this documentation hasn’t been updated yet.
Gumloop offers a range of AI models to cater to various needs and complexities of tasks. Our AI models are divided into three main categories based on their capabilities and credit costs.

​
Model Categories Overview

Expert AI Models

Most capable models for complex tasks requiring advanced reasoning30 credits per call

Advanced AI Models

Strong performance for most business use cases with balanced speed20 credits per call

Standard AI Models

Cost-effective solutions for common tasks with faster processing2 credits per call

​
Expert AI Models

Best suited for critical tasks where accuracy and reasoning quality are paramount.

​
Available Models

  • GPT-5.4
  • GPT-5.3 Codex
  • GPT-5.2
  • GPT-5.2 Codex
  • GPT-5.1
  • GPT-5
  • Claude 4.6 Opus
  • Claude 4.5 Opus
  • Claude 4.1 Opus
  • Gemini 3.1 Pro
  • Grok 4
  • Qwen3.5 397B
  • Kimi K2.5

​
When to Use Expert Models

  • Higher accuracy on complex and specialized tasks
  • Better reasoning capabilities for nuanced problems
  • Lower rate of inaccuracies in generated content
  • Strong performance on specialized and technical requests
  • Slower response times compared to other models
  • Higher credit cost (30 credits per call)

​
Advanced AI Models

Appropriate for most business use cases that require good performance without the highest credit cost.

​
Available Models

  • Claude 4.6 Sonnet
  • Claude 4.5 Sonnet
  • Claude 4 Sonnet
  • Gemini 2.5 Pro
  • Perplexity Sonar Reasoning Pro
  • Perplexity Sonar Pro
  • LLaMA 3 405B Instruct
  • Grok 3

​
When to Use Advanced Models

  • Good performance on complex tasks
  • Effective reasoning and content generation
  • Lower error rate than standard models
  • Suitable for most business use cases
  • Moderate response times
  • Higher credit cost (20 credits per call)

​
Standard AI Models

Suitable for general tasks like basic summarization, simple content generation, and straightforward analysis.

​
Available Models

  • GPT-5.4 Mini
  • GPT-5.4 Nano
  • GPT-5 Mini
  • GPT-5 Nano
  • Claude 4.5 Haiku
  • Gemini 3 Flash
  • Gemini 2.5 Flash
  • Perplexity Sonar
  • LLaMA 3 70B
  • Grok 3 Mini
  • DeepSeek V3
  • DeepSeek R1
  • Mixtral 8x7B 32K

​
When to Use Standard Models

  • Lower credit cost (2 credits per call)
  • Faster processing for quicker workflow execution
  • Sufficient quality for many common use cases
  • Less reliable on complex or ambiguous tasks
  • Higher error rate compared to Advanced or Expert models
  • May struggle with nuanced or technical requests

​
Specialized Models

Azure OpenAI

Connect your own Azure OpenAI service to use your own deploymentRequires Credentials

Auto-Select

Automatically selects an appropriate model based on the taskVariable cost

Deep Research

Perplexity Sonar Deep Research for in-depth web research tasksResearch tier

​
Vision Models

Vision models are designed to analyze and generate visual content. These models can perform tasks such as image recognition, analysis, and object detection.

​
Models Available for Image Analysis

The “Analyze Image” node offers vision-capable models across all three tiers:
  • GPT-5.4
  • GPT-5.3 Codex
  • GPT-5.2
  • GPT-5.1
  • GPT-5
  • Claude 4.6 Opus
  • Claude 4.5 Opus
  • Claude 4.1 Opus
  • Gemini 3.1 Pro
  • Grok 4

​
Using Your Own API Keys (BYOK)

Provide your own API keys for AI services to significantly reduce credit costs. The discount depends on whether you’re using workflows or agents:
Automation TypeWithout BYOKWith BYOKSavings
Workflow AI Nodes2-30 credits per call (varies by model tier)1 credit per callUp to 97%
Agent AI Model CreditsFull token-based cost50% off token-based cost50%
Tool credits and workflow base costs are not affected by BYOK — the discount applies only to AI model usage.

For Workflows

  • Use free native nodes when possible
  • Provide your own API keys (reduces AI costs to 1 credit)
  • Batch operations to reduce workflow runs
  • Avoid unnecessary enrichment lookups

For Agents

  • Provide your own API key (50% off AI model credits)
  • Choose appropriate models (don’t use Expert for simple tasks)
  • Start new conversations for different topics
  • Limit tools to only what’s needed
  • Write clear, specific prompts
  • Optimize workflows that agents call
Requirements:
  • Pro plan or higher
  • Your own OpenAI, Anthropic, or Google AI account
Setup: Add your API key in Credentials Settings