In a recent LinkedIn post, Ruben Hassid offers a practical framework for selecting the most appropriate Claude AI model for specific tasks, aiming to help users optimize performance and manage costs effectively. Hassid breaks down the decision-making process into simple steps, emphasizing that not all tasks require the most advanced AI.
Hassid introduces his tiered approach by posing a fundamental question: does the task demand a complex answer? He suggests that for straightforward needs, lighter models suffice, while more intricate problems warrant more powerful options. As Ruben Hassid outlines:
“No: use Haiku 4.5 or Sonnet 5. ☑️ Yes: use Opus 4.8 or Fable 5.”
Optimizing for Speed and Simplicity
For users prioritizing speed and minimal token usage, Hassid recommends Claude Haiku 4.5. This model is presented as a lightweight, token-efficient option ideal for quick chats or tasks where brevity is key. Hassid suggests enabling web search for Haiku and using features like “Plan in Chat, build in Cowork.” For everyday tasks that are not overly complex, Claude Sonnet 5 is presented as the go-to model, offering a balance of speed and capability. Hassid also highlights Sonnet’s integration capabilities, noting its compatibility with a wide array of connectors for tools like Slack, Google Drive, and Notion.
In his post, Hassid provides specific prompt examples tailored to these models. For Haiku, he suggests:
“I want [desired result] with [constraints]. Ask me questions using AskUserQuestion before you start.”
And for Sonnet:
“You are a [role]. [Task] this [input]. Keep it under [length]. Tone: [casual/formal]. No preamble – just the output.”
Leveraging Advanced Models for Complex Challenges
When tackling more demanding projects, Ruben Hassid directs users towards Claude Opus 4.8. This model is positioned for “deep-work” and multi-step reasoning tasks. Hassid advises users to utilize features like “Cowork” and set “Effort on High” for optimal results. He also points to the utility of Claude Skills within Opus, such as specific commands for LinkedIn or Excel-like functions.
For the most ambitious and intellectually intensive tasks, Hassid introduces Claude Fable 5 as the “smartest model by Claude.” Fable 5 is recommended for deep research and analytical decision-making. He provides a prompt structure for Fable 5 designed to elicit thorough reasoning:
“Here is my goal: [goal]. Here are my constraints: [constraints]. Think through the tradeoffs before answering, propose 2–3 approaches, and recommend one with your reasoning.”
Cost Considerations and Strategic Usage
Crucially, Hassid issues a caution regarding Fable 5, noting that it incurs additional costs beyond a certain usage threshold after July 12th and that only a small percentage of tasks truly necessitate its power. He suggests a strategic approach: use Fable 5 for initial strategy or complex analysis (1-2 turns) and then transition to Opus for the execution of the task. This method, according to Hassid, helps manage costs, as Claude models re-process the entire conversation history with each turn, making longer interactions with the most advanced models potentially expensive.
Hassid also shares a valuable resource he developed: a “/fable-prompter” skill designed to refine messy prompts into Fable-worthy ones. This skill, along with over 26 other Claude Skills, is available through his “how-to-ai.guide” website. He encourages users to sign up via email to access his library of AI tools.
Ultimately, Ruben Hassid’s post serves as a valuable guide for anyone looking to harness the power of Claude AI more efficiently. By providing a clear, step-by-step methodology and practical prompt examples, Hassid empowers users to select the right model, optimize their interactions, and avoid unnecessary expenditure on AI processing.
📝 About This Content
This article is based on insights shared by Ruben Hassid on LinkedIn.
📅 Originally posted on July 10, 2026 | View original post on LinkedIn →