DailyAi.Work

Pick the Right AI Model for a Task

You are an AI infrastructure and model selection expert.

Category
AI & Prompt Engineering
Output
Text
Fields
4
  • prompt-engineering
  • workflow-design
  • ai-communication
  • output-optimization
  • critical-evaluation

Fill in the details

Describe the specific task or workflow you want to run on an AI model.

Key priorities

Select the factors that matter most for your use case, such as accuracy, latency, or budget.

Indicate how often this task will run to help determine the right architecture and pricing tier.

List any technical limits, budget caps, compliance rules, or integration requirements, or leave blank if none apply.

Prompt

You are an AI infrastructure and model selection expert. Analyze the user’s task and recommend the most suitable AI model or routing strategy based on their priorities, expected volume, and constraints. Follow these steps: 1. **Task Analysis**: Identify what the task actually demands from a model (e.g., reasoning depth, creativity, structured output, long-context handling, latency sensitivity, or deterministic formatting). 2. **Priority Tradeoffs**: Evaluate how different model tiers balance the user’s stated priorities. Explain which factors will inevitably be compromised and why. 3. **Model Recommendation**: Suggest 2–3 capability tiers or architectures that fit the use case (e.g., frontier reasoning models, optimized mid-tier models, open-source fine-tunes, or specialized API endpoints). Specify when to use each tier. 4. **Routing Strategy**: Assess whether a single-model approach or a dynamic routing strategy (e.g., fallback chains, task-specific routing, or hybrid setups) makes more sense for this workload. 5. **Testing Framework**: Provide a practical, step-by-step method to compare candidate models using the user’s real inputs, including evaluation metrics and success criteria. 6. **Cost Optimization**: List actionable cost-saving tactics that preserve quality (e.g., caching, prompt compression, tiered routing, quantization, or batch processing). Constraints & Rules: - Base all recommendations strictly on the provided Task or use case, Key priorities, Expected volume, and Constraints & limits. - Do not invent model capabilities, pricing, latency figures, or benchmarks. If you lack specific data, state it clearly. - If any critical information is missing or ambiguous, list it in a short “Open questions” section at the end instead of guessing. - Keep recommendations practical and capability-focused. Avoid marketing hype or brand-name chasing unless directly relevant to architecture choices. - Format the output with clear headings for each of the six steps, using bullet points for readability.

How to use this prompt

  1. Fill in the fields above. Required ones are marked with *.
  2. Check the prompt preview: your answers appear highlighted in place.
  3. Select Copy prompt and paste it into ChatGPT, Claude, Gemini or any AI assistant.

More AI & Prompt Engineering prompts

See all AI & Prompt Engineering prompts