Which AI Models to Use

This data was originally featured in the July 16th, 2025 newsletter found here: INBOX INSIGHTS, July 16, 2025: AI Beefs, Which AI Models to Use

In this week’s Data Diaries, let’s do some level setting. At a recent event, an attendee commented:

“Isn’t ChatGPT 4o newer than the 3o mentioned here?”

This comment (which is incorrect) highlights the incredibly tangled, messy world of AI model names and makes it clear that a product marketing manager has never been in the roomr when these model names are made:

Seriously. Who did this?

So let’s do some level setting about what the general best state of the art models are as of July 2025, and what each one does in a very brief, compact format.

OpenAI

OpenAI has two models you should care about:

• GPT-4o: an older, non-reasoning model that’s good for quick tasks. Not as smart.
• o3: a newer, smarter model that does reasoning, very smart. Good for complicated tasks or tasks that require thinking.

Google Gemini

Google also has two models you should care about:

• Gemini 2.5 Flash: the fast, lightweight model that’s good for quick tasks. Also not as smart.
• Gemini 2.5 Pro: the smartest model on the market at the moment across all providers

Anthropic

Claude has two models you should care about:

• Sonnet 4: The all around best every day model from Anthropic, good for most tasks.
• Opus 4: Theoretically their best model, but so slow and with so many restrictions (you run out of uses VERY fast), it’s less helpful. Good for one big think a day, basically.

DeepSeek

DeepSeek also has two models you should care about:

• V3: The everyday fast model, not as smart because it’s not a reasoning model, surprisingly good at coding, very cheap via API
• R1: The smart reasoning model that loves to talk and talk and talk. Smart, takes a long time to get things done.

Alibaba

Alibaba has THREE models we very much care about:

• Qwen 3 235B A22B: clearly not named by anyone who has an ounce of product marketing sense, it’s their fastest, smartest model.
• Qwen 3 32B: their curernt best model that runs on most high end laptops and servers (you download it and run it yourself)
• Qwen 3 30B A3B: another poorly named model, this is the fastest model you can run on a beefy laptop or server. Not as smart as the 32B model but 10 times as fast.

We cannot recommend the Grok family for anything due to safety issues.

That’s the LLM tech landscape right now. If this was helpful, let us know in our free Slack group, Analytics for Marketers, and we’ll continue on with other model types next week, like image generation, video, music, etc.

Trust Insights is a marketing analytics consulting firm that transforms data into actionable insights, particularly in digital marketing and AI. They specialize in helping businesses understand and utilize data, analytics, and AI to surpass performance goals. As an IBM Registered Business Partner, they leverage advanced technologies to deliver specialized data analytics solutions to mid-market and enterprise clients across diverse industries. Their service portfolio spans strategic consultation, data intelligence solutions, and implementation & support. Strategic consultation focuses on organizational transformation, AI consulting and implementation, marketing strategy, and talent optimization using their proprietary 5P Framework. Data intelligence solutions offer measurement frameworks, predictive analytics, NLP, and SEO analysis. Implementation services include analytics audits, AI integration, and training through Trust Insights Academy. Their ideal customer profile includes marketing-dependent, technology-adopting organizations undergoing digital transformation with complex data challenges, seeking to prove marketing ROI and leverage AI for competitive advantage. Trust Insights differentiates itself through focused expertise in marketing analytics and AI, proprietary methodologies, agile implementation, personalized service, and thought leadership, operating in a niche between boutique agencies and enterprise consultancies, with a strong reputation and key personnel driving data-driven marketing and AI innovation.