What AI Models Does August Use

What AI Models Does August Use

Overview

August does not rely on a single large language model. Instead, it uses a multi model system that selects the best model or combination of models for each task. This gives you both flexibility and strong performance. August routes each request intelligently and then combines model outputs to produce drafting, analysis, review, or extraction results that match your goal.

Below is an overview of the types of models August uses and how the system selects them.

Models in Our Suite

August supports several types of advanced models. Each type is suited to different work:

• Large scale models designed for long form drafting and deeper reasoning
• Models optimized for analysis, summarization, clause extraction, and structured workflows
• High throughput models built for fast processing of large document sets such as contracts, policies, or pleadings

The multi model approach ensures no single model has to perform every kind of task. Instead, August selects the right tool for each part of the job.

How Model Selection Works

When you submit a request such as drafting a memo, reviewing a contract, running a workflow, or summarizing a regulation, August:

1. Breaks the request into clear steps such as analysis, extraction, drafting, or summarizing
2. Selects the best model for each step based on reasoning needs, document length, or task complexity
3. Runs the selected model or models
4. Combines the outputs into a single final result such as a draft, table, summary, or redline

This approach increases both accuracy and efficiency. You get detailed writing when needed and fast structured output when handling large volumes.

Auto Mode

Most users do not need to pick a model manually. Auto Mode handles model selection for you.

Agentic Mode

Agentic mode is used when a task requires several steps of reasoning. Choose this mode when you need:

• Multi step logical analysis
• Tasks that require planning, checking, or verification
• Work that depends on multiple documents or conditions
• Structured problem solving such as comparing conflicting clauses

Agentic mode is helpful when the work resembles what a junior associate would perform across several steps rather than one direct answer.

Deep Research

Deep Research is used when you need long form analysis or synthesis across multiple sources. Choose this mode when you want:

• Thorough research memos
• Multi document comparisons
• Detailed regulatory summaries
• Comprehensive factual reviews

Deep Research is ideal for long form work where depth matters more than speed.

Manual Model Selection

If enabled, a model selector allows users to choose a model directly for specialized tasks or internal preferences. Most tasks should still rely on Auto Mode unless you have a specific reason to override it.

Our Model Evaluation Process

Before any model is used in August, it goes through a detailed evaluation process that includes:

• Testing on legal benchmarks to measure reasoning, citation accuracy, contract analysis, and drafting quality
• Product testing to confirm stability and performance on real world document sets
• Human legal review to evaluate clarity, quality, and alignment with legal work

Models are reevaluated regularly to ensure reliability over time.

Why Multi Model Matters

• Flexibility across tasks: Some work requires deep reasoning. Other work requires speed. Using multiple models gives you both.
• Resilience and performance: If one model struggles, another can support the task. This reduces errors and improves reliability.
• Scalability: For large projects, August can process documents faster by using lighter models where appropriate.
• Control: Firms can allow Auto Mode, enable model selection, or choose a consistent model for certain matters.

What Users Should Know

• You rarely need to choose a model. Auto Mode picks the best option for you.
• If your workspace allows manual selection, advanced users can choose a model for specific tasks.
• Output quality will vary depending on model type, task complexity, and document volume.
• Models evolve over time. Performance improves as new model versions are introduced and older ones are retired.