• The Brainyacts
  • Posts
  • 263 | GPT, Claude, Gemini, Grok... Wait, Which One Do I Use Again?

263 | GPT, Claude, Gemini, Grok... Wait, Which One Do I Use Again?

Brainyacts #263

Hello to all 7500+ of you from around the globe. It’s been a minute.

One thing keeps coming up in conversations—with colleagues, students, and legal professionals alike: “Which AI model should I actually be using?”

Fair question. There are a lot of models out there now. GPT-4, GPT-4.5, Claude 3, Gemini 1.5, Mistral… The names aren’t helpful. The differences aren’t obvious. And unless you’re living inside the AI news cycle (which I don’t recommend), you’re probably using whatever tool you started with out of habit.

So this edition is simple: a quick, practical guide to the major generative AI models available in 2025 so far. What they’re good at, what to use them for, and where they might fit into your legal work—from document summarization to client communication to research support.

Let’s dive in.

A Practical Guide to the Major Generative AI Models

1. OpenAI Models

OpenAI is a leader in AI, and its models power tools like ChatGPT. Here’s a look at their current offerings for 2025. Sign up for ChatGPT at chatgpt.com. Start with o4-mini (free) or upgrade to Plus ($20/month) for GPT-4o and o3.

GPT-4o

  • What It Is: GPT-4o is OpenAI’s most advanced all-purpose model, replacing GPT-4 (which is no longer available after April 30, 2025). It’s fast, handles text and images, and is great for a wide range of tasks. Available with a ChatGPT Plus subscription ($20/month).

  • Strengths:

    • Understands and generates clear, human-like text.

    • Can analyze images (e.g., scanned contracts or handwritten notes).

    • Fast responses, ideal for real-time tasks like client chats.

    • Fewer errors (hallucinations) than older models.

  • Best Legal Use Cases:

    • Drafting Documents: Write professional emails, letters, or basic contracts. Example: Ask GPT-4o to draft a client retainer agreement based on your firm’s template.

    • Summarizing Texts: Summarize long case law or deposition transcripts in seconds. Example: “Summarize this 20-page judgment in 200 words.”

    • Client Communication: Generate polite, tailored responses to client inquiries. Example: “Write a friendly email explaining a delay in case progress.”

    • Image Analysis: Extract text from scanned legal documents or analyze diagrams in patents.

  • Responsible Use Tip: Always check AI-drafted documents for legal accuracy and jurisdiction-specific details. GPT-4o is versatile but may not know niche local laws.

GPT-4.5 Research Preview

  • What It Is: GPT-4.5 (codenamed Orion) is OpenAI’s newest, largest model, available only to ChatGPT Pro subscribers ($200/month) as a research preview. It’s designed for complex tasks and has deeper knowledge.

  • Strengths:

    • Excellent at understanding context and intent, making it feel more “thoughtful.”

    • Stronger at writing, problem-solving, and analyzing complex texts.

    • Reduced errors compared to GPT-4o, but it’s slower and more expensive.

  • Best Legal Use Cases:

    • Complex Legal Writing: Draft detailed legal memos or early draft appellate briefs. Example: “Write a memo analyzing the implications of a new privacy law.”

    • Deep Analysis: Break down intricate regulatory texts or multi-party contracts. Example: “Explain the risks in this 50-page merger agreement.”

    • Persuasive Communication: Craft compelling arguments for negotiations or mediation. Example: “Write a letter persuading opposing counsel to settle.”

  • Responsible Use Tip: GPT-4.5 is powerful but costly. Use it for high-stakes tasks where precision matters, and always verify outputs for factual accuracy, as it’s still in preview.

o3

  • What It Is: o3 is OpenAI’s reasoning-focused model, designed to “think” through problems step-by-step. Available with ChatGPT Plus ($20/month) or Pro ($200/month) plans.

  • Strengths:

    • Excels at logical reasoning, math, and structured tasks.

    • Breaks down problems into clear steps, making it easier to follow.

    • Ideal for tasks requiring precision and reliability.

  • Best Legal Use Cases:

    • Legal Reasoning: Analyze hypothetical scenarios or case outcomes. Example: “What are the likely outcomes of this contract dispute based on these facts?”

    • Structured Summaries: Create detailed outlines of legal arguments or case briefs. Example: “Outline the key points of this statute in bullet points.”

    • Compliance Checks: Evaluate documents for regulatory compliance. Example: “Check if this policy aligns with GDPR requirements.”

  • Responsible Use Tip: o3 is great for logic-driven tasks but less creative than GPT-4o. Use it when you need clear, step-by-step analysis, and double-check for legal nuances.

o4-mini

  • What It Is: o4-mini is a smaller, faster, and cheaper version of GPT-4o, ideal for quick tasks. Available with ChatGPT Plus ($20/month) or free with usage limits.

  • Strengths:

    • Lightning-fast responses, perfect for simple queries.

    • Cost-effective for high-volume tasks.

    • Still handles text and images but less powerful than GPT-4o.

  • Best Legal Use Cases:

    • Quick Drafts: Write short emails or meeting agendas. Example: “Draft a 100-word email scheduling a client meeting.”

    • Basic Summaries: Summarize short documents or articles. Example: “Summarize this 2-page legal blog post.”

    • Document Formatting: Reformat text for clarity. Example: “Convert this list of case points into a table.”

  • Responsible Use Tip: o4-mini is great for speed but may oversimplify complex legal issues. Use it for routine tasks and review outputs carefully. 

2. Google Gemini Models

Google’s Gemini models are known for speed, integration with Google tools, and handling large amounts of text. They’re great for legal professionals already using Google Workspace. Visit gemini.google.com. Use 2.0 Flash for free or get Gemini Advanced ($19.99/month) for 2.5 Flash/Pro and Deep Research.

Gemini 2.0 Flash

  • What It Is: A fast, lightweight model for quick tasks, available with a free Google account or Gemini Advanced ($19.99/month) for higher limits.

  • Strengths:

    • Super-fast responses, ideal for time-sensitive work.

    • Integrates with Google apps like Docs and Sheets.

    • Handles text and basic image analysis.

  • Best Legal Use Cases:

    • Quick Research: Find definitions or basic legal concepts. Example: “Explain ‘res ipsa loquitur’ in 100 words.”

    • Document Editing: Edit drafts in Google Docs. Example: “Suggest edits to make this contract clause clearer.”

    • Task Automation: Generate lists or schedules. Example: “Create a checklist for filing a motion.”

  • Responsible Use Tip: Gemini 2.0 Flash is fast but less nuanced. Use it for straightforward tasks and verify outputs for depth and accuracy.

Gemini 2.5 Flash

  • What It Is: A step up from 2.0 Flash, with better performance and a larger context window (ability to process more text at once). Available with Gemini Advanced ($19.99/month).

  • Strengths:

    • Handles longer documents and more complex queries.

    • Strong at summarizing and organizing information.

    • Still fast and cost-efficient.

  • Best Legal Use Cases:

    • Long Document Summaries: Summarize multi-page filings or reports. Example: “Summarize this 30-page discovery document.”

    • Data Organization: Organize case notes or evidence lists. Example: “Sort these case facts into a timeline.”

    • Multimodal Tasks: Analyze text and images together, like annotated legal exhibits.

  • Responsible Use Tip: Great for processing large texts, but ensure outputs align with legal standards, as it may miss subtle legal distinctions.

Gemini 2.5 Pro

  • What It Is: Google’s most advanced model, designed for complex tasks and coding. Available with Gemini Advanced ($19.99/month).

  • Strengths:

    • Excels at coding, web app development, and handling massive documents (2 million token context window).

    • Strong reasoning and multimodal capabilities (text, images, video).

    • Integrates with Google’s ecosystem for seamless workflows.

  • Best Legal Use Cases:

    • Complex Document Analysis: Process entire case files or statutes. Example: “Analyze this 100-page case file for key arguments.”

    • Legal Tech Development: Build simple tools, like a contract review script. Example: “Write a script to flag risky clauses in contracts.”

    • Video Summaries: Summarize deposition videos uploaded to Google Drive. Example: “Summarize this 1-hour deposition video.”

  • Responsible Use Tip: Powerful for tech-savvy firms, but coding and video features require technical oversight. Always validate legal outputs.

Deep Research (Gemini)

  • What It Is: A feature in Gemini Advanced that lets the model search the web and analyze multiple sources to provide detailed answers. Think of it as a research assistant compiling a report.

  • Strengths:

    • Pulls from diverse online sources for comprehensive answers.

    • Summarizes findings clearly, citing sources.

    • Ideal for broad or current topics.

  • Best Legal Use Cases:

    • Preliminary Research: Explore recent case law or regulations. Example: “Research recent U.S. Supreme Court rulings on data privacy.”

    • Market Analysis: Understand trends in legal tech. Example: “What are the latest AI tools for law firms?”

    • Fact-Checking: Verify legal claims or news. Example: “Is this article about a new law accurate?”

  • Responsible Use Tip: Deep Research is great for starting points but not a substitute for Westlaw or LexisNexis. Always cross-check with primary legal sources, as web data may be incomplete or biased.

Personalization (Gemini)

  • What It Is: A feature that lets Gemini learn from your Google Workspace data (e.g., emails, Docs) to tailor responses to your preferences or firm’s style.

  • Strengths:

    • Adapts to your writing style or firm’s tone.

    • Saves time by aligning outputs with your needs.

    • Enhances integration with Google tools.

3. xAI Grok

xAI Grok: Access Grok at x.ai or via X. Free with limits, or get X Premium ($50/month) for full access.

Grok 3

  • What It Is: Created by xAI, Grok 3 is a fast, witty model integrated with X (formerly Twitter) for real-time information. Available free with limits or with X Premium ($50/month) for higher usage.

  • Strengths:

    • Accesses real-time X posts, great for current events or trends.

    • Strong at math, science, and coding, but also conversational.

    • Humorous, human-like tone with fewer strict guardrails.

    • Includes image generation (Aurora) and analysis.

  • Best Legal Use Cases:

    • Current Legal Trends: Track discussions on legal topics via X. Example: “What are lawyers on X saying about AI in law?”

    • Quick Coding: Build simple legal tech tools. Example: “Write a script to organize case files by date.”

    • Image-Based Tasks: Analyze or generate visuals, like court exhibit diagrams. Example: “Generate a timeline graphic for this case.”

    • Brainstorming: Generate creative ideas for case strategies. Example: “Suggest negotiation tactics for a contract dispute.”

  • Responsible Use Tip: Grok’s real-time data can be noisy or biased. Verify X-based insights with reputable sources, and ensure image outputs comply with ethical standards.

Deep Research (Grok)

  • What It Is: Grok’s “DeepSearch” feature searches the web and X posts to provide in-depth answers, similar to Gemini’s Deep Research.

  • Strengths:

    • Combines web and X data for unique perspectives.

    • Fast and conversational, with a focus on recent information.

    • Free to use (with limits).

  • Best Legal Use Cases:

    • Public Sentiment Analysis: Gauge opinions on legal issues. Example: “What’s the public reaction to this new law on X?”

    • News Monitoring: Stay updated on legal developments. Example: “Find recent articles on AI ethics in law.”

    • Exploratory Research: Get a quick overview of emerging topics. Example: “What’s new in legal tech for 2025?”

  • Responsible Use Tip: DeepSearch is useful for trends but not authoritative. Cross-check with legal databases, and be cautious of unverified X posts.

4. Anthropic Claude Models

Anthropic’s Claude models are known for safety, ethical design, and structured outputs, making them ideal for regulated fields like law. Anthropic Claude: Try Claude at claude.ai. Free for Haiku with limits, or get Pro ($20/month) for Sonnet and Opus.

Claude 3.7 Sonnet

  • What It Is: Anthropic’s most advanced model, blending quick responses with deep reasoning. Available with a Claude Pro subscription ($20/month).

  • Strengths:

    • Excels at structured writing, coding, and logical analysis.

    • “Extended thinking” mode for complex problem-solving.

    • Low error rates and a neutral, professional tone.

    • Can’t generate images but analyzes them well.

  • Best Legal Use Cases:

    • Structured Documents: Draft organized legal briefs or policies. Example: “Write a compliance policy for data protection.”

    • Technical Analysis: Analyze complex legal texts or contracts. Example: “Identify ambiguities in this lease agreement.”

    • Coding for Legal Tech: Build tools like automated document checkers. Example: “Create a script to highlight key terms in contracts.”

    • Client Support: Generate detailed, professional responses. Example: “Answer this client query about trademark law.”

  • Responsible Use Tip: Claude’s safety focus makes it great for sensitive tasks, but it lacks web access. Supplement with manual research for current laws.

Claude 3.5 Haiku

  • What It Is: A fast, cost-effective model for quick tasks, matching the power of older high-end models. Available with Claude Pro ($20/month) or free with limits.

  • Strengths:

    • Blazing-fast responses, ideal for high-volume work.

    • Strong at translation, editing, and data processing.

    • Compact but still handles complex tasks well.

  • Best Legal Use Cases:

    • Translation: Translate legal documents or client communications. Example: “Translate this contract into Spanish.”

    • Editorial Tasks: Polish drafts or rephrase text. Example: “Make this legal memo more concise.”

    • Data Processing: Extract key points from unstructured texts. Example: “List the main obligations in this regulation.”

  • Responsible Use Tip: Haiku is efficient but less nuanced than Sonnet. Use it for routine tasks, and review outputs for depth in complex cases.

Claude 3 Opus

  • What It Is: Anthropic’s most powerful model for complex tasks, designed for deep analysis. Available with Claude Pro ($20/month).

  • Strengths:

    • Handles extensive context, perfect for long conversations or documents.

    • Excels at nuanced, high-stakes tasks requiring precision.

    • Strong vision capabilities for analyzing images.

  • Best Legal Use Cases:

    • In-Depth Research Support: Analyze large datasets or case histories. Example: “Compare these five cases on intellectual property.”

    • Work Automation: Streamline repetitive tasks like document categorization. Example: “Sort these emails into ‘urgent’ and ‘non-urgent.’”

    • Vision Tasks: Interpret scanned legal forms or diagrams. Example: “Extract text from this handwritten affidavit.”

  • Responsible Use Tip: Opus is resource-intensive, so reserve it for complex tasks. Always verify outputs, especially for niche legal areas.

Responsible AI Use in Law: Key Reminders

AI can be a game-changer, but it’s not perfect. Here are tips to use these models responsibly:

  1. Verify Outputs: AI can make mistakes or miss legal nuances. Always cross-check with primary sources like case law or statutes.

  2. Protect Client Data: Avoid inputting sensitive client information unless you’re using a secure, compliant platform. Check your firm’s AI policies.

  3. Understand Limitations: Web access varies across models and is not uniform or consistent even within a model. Use the right tool for the job.

  4. Stay Ethical: Ensure AI use complies with professional conduct rules and any employer policies. Don’t rely on AI for final legal advice without human review.

  5. Start Small: If you’re new, try free or low-cost models like o4-mini or Gemini 2.0 Flash to build confidence before investing in subscriptions.

Final Thoughts

Start small, stay curious, and always review before relying. As these tools evolve, so will our understanding of how they best serve the legal profession. I’ll continue to share updates as things shift.

In the meantime, if you’ve tried any of these models—or built something interesting with them—I’d love to hear about it. Just hit reply.

Until next time,

Josh

To read previous editions, click here.

Was this newsletter useful? Help me to improve!

With your feedback, I can improve the letter. Click on a link to vote:

Login or Subscribe to participate in polls.

Who is the author, Josh Kubicki?

I am a lawyer, entrepreneur, and teacher. Not a theorist, I am an applied researcher and former Chief Strategy Officer, recognized by Fast Company and Bloomberg Law for my work. Through this newsletter, I offer you pragmatic insights into leveraging AI to inform and improve your daily life in legal services.

DISCLAIMER: None of this is legal advice. This newsletter is strictly educational and is not legal advice or a solicitation to buy or sell any assets or to make any legal decisions. Please /be careful and do your own research.