Models | Clear Ideas Documentation

Foundational Models

Clear Ideas integrates advanced language models (often called foundational models) from providers including OpenAI, Anthropic, Google, Cohere, XAI, and DeepSeek. This document describes how these models are used within the Clear Ideas platform, along with considerations for choosing the appropriate model.

Foundational models are AI models trained on large datasets that can generate text based on a user-provided prompt. They are designed to perform a wide range of tasks, such as:

Factual Q&A
Summarization
Creative writing
Technical/code-related queries
Data analysis

Clear Ideas offers multiple models, each with different strengths, performance trade-offs, and associated costs.

Supported Models

Intelligent

Selects the most appropriate model for the task based on task type, reasoning level required, desired response length, target audience, creativity needs, and time sensitivity. Optimizes for cost efficiency when capabilities are equal.

GPT-4o-mini

Open AI's fast, affordable model for focused tasks. Excellent for summaries, factual Q&A, and straightforward tasks requiring simple reasoning. Delivers brief, concise responses for general audiences with high time sensitivity. Cost-efficient choice for most everyday inquiries.

GPT-4o

OpenAI's versatile, high-intelligence flagship model. Excels in interactive conversations with balanced reasoning capabilities. Ideal for content creation, creative writing, code/technical tasks, and ethical/philosophical discussions. Best for professional audiences requiring medium-length responses with moderate time sensitivity.

OpenAI o3-mini

OpenAI's fast and affordable reasoning model trained with reinforcement learning to perform complex reasoning. Ideal for moderate reasoning tasks, professional audiences, and scenarios with medium time sensitivity. Best for direct responses with basic reasoning and low creativity needs.

OpenAI o3-mini with High Reasoning Effort

OpenAI's fast and affordable reasoning model trained with reinforcement learning to perform complex reasoning with a high reasoning effort. Excels at detailed analysis and comprehensive answers for complex reasoning tasks. Ideal for technical/expert audiences when time sensitivity is low but thorough analysis is required.

Gemini 2.0 Flash

Google's most capable multi-modal model with great performance across all tasks. Excels at factual Q&A and summarization with simple reasoning. Ideal for general audiences requiring brief responses with high time sensitivity. The most cost-efficient option when low to moderate reasoning is sufficient.

Claude 3.7 Sonnet

Anthropic's most intelligent model. Excels at complex reasoning, creative content generation, code/technical tasks, data analysis, and ethical/philosophical discussions. Ideal for technical/expert audiences requiring comprehensive responses. Best when creativity needs are high and time sensitivity is low.

Claude 3.7 Sonnet with Enhanced Reasoning

Anthropic's most intelligent model with enhanced reasoning. The highest reasoning capability for advanced problem-solving, mathematical/logical proofs, and in-depth analysis. Ideal for academic/research audiences when accuracy is prioritized over speed. Best for scenarios requiring the most rigorous and thorough analysis possible.

Command A

Cohere's Command A is a sophisticated large language model designed to assist users by providing thorough, accurate, and contextually relevant responses to a wide range of queries, while maintaining high efficiency in processing and delivering information, ensuring quick and reliable assistance.

Reasoning vs. Conversational Models

Clear Ideas offers models with a primary emphasis on reasoning or conversational capabilities. While there can be overlap, understanding each focus can help match the model to the task:

Reasoning Models: Designed to handle more complex analytical tasks. These models are used when the goal is to perform in-depth analysis, solve technical problems, or produce detailed explanations. Reasoning-focused models often require more computational resources and can be more time-intensive, but they excel at:
- Complex problem-solving
- Logical or mathematical tasks
- Technical discussions and code assistance
- Data-driven analysis or research
Conversational Models: Tailored primarily for natural, back-and-forth interactions. They are useful for human-like dialogue, making them well suited for:
- Q&A sessions or short clarifications
- Casual engagement or interactive scenarios
- Iterative brainstorming discussions
- High-level summaries or succinct explanations

When choosing between a reasoning or conversational model, consider the depth and complexity of your query. If the task involves sophisticated analysis, a reasoning model can be more appropriate. If you need an approachable dialogue or iterative exchange, a conversational model may be preferable.

Intelligent Model Selection

Clear Ideas provides an Intelligent auto-selector that evaluates multiple factors—such as task complexity, desired detail level, audience type, and cost constraints—to pick the most suitable model. If multiple models appear equally suitable, the selection process favors the lower-cost option. This approach reduces the need for manual model comparison.

Model Selection Considerations

Task Complexity

Lower complexity (summarization, short Q&A):
Higher complexity (in-depth analysis, technical tasks):

Response Length

Brief responses
Extended output

Audience

General audience
Expert audience

Time Sensitivity

High time sensitivity
Lower time sensitivity

Creativity

Lower creativity needed
Higher creativity needed