Skip to Content
The App Store for AI Agents - Run, create, and monetize AI glyphs
TopicsModel Tiers

Model Tiers

Understanding AI model tiers for prompt glyphs.

Tier Overview

TierPriceSpeedQualityBest For
Fast$0.022-5sGoodSimple tasks
Balanced$0.055-15sGreatMost use cases
Reasoning$0.1015-60sBestComplex analysis

Fast Tier

Price: $0.02 per run

Models

  • GPT-4o Mini (OpenAI)
  • Claude 3 Haiku (Anthropic)
  • Llama 3.1 8B (Meta)
  • Gemini Flash (Google)
  • Mistral 7B (Mistral)

Characteristics

  • Fastest response times
  • Lower cost
  • Good for straightforward tasks
  • May miss nuance

Best For

  • Simple formatting
  • Quick translations
  • Basic summaries
  • Short answers
  • High-volume tasks

Example Use

“Generate 5 token name ideas for a DeFi project”

Balanced Tier

Price: $0.05 per run

Models

  • GPT-4o (OpenAI)
  • Claude 3.5 Sonnet (Anthropic)
  • Llama 3.1 70B (Meta)
  • Gemini Pro (Google)
  • Mistral Large (Mistral)

Characteristics

  • Optimal quality/speed balance
  • Handles complex tasks well
  • Good reasoning ability
  • Most popular choice

Best For

  • Code analysis
  • Content generation
  • Technical writing
  • Strategy development
  • Most general tasks

Example Use

“Analyze this smart contract for security vulnerabilities and provide recommendations”

Reasoning Tier

Price: $0.10 per run

Models

  • o1-preview (OpenAI)
  • o1-mini (OpenAI)
  • Claude 3 Opus (Anthropic)
  • Llama 3.1 405B (Meta)
  • DeepSeek V3 (DeepSeek)

Characteristics

  • Highest quality output
  • Deep reasoning ability
  • Handles complex analysis
  • Longer response times

Best For

  • Complex problem solving
  • Multi-step reasoning
  • Deep technical analysis
  • Critical decisions
  • Research tasks

Example Use

“Perform a comprehensive security audit of this DeFi protocol, analyzing economic attack vectors, smart contract vulnerabilities, and governance risks”

Choosing a Tier

Consider Your Task

Task TypeRecommended
Quick lookupFast
FormattingFast
General contentBalanced
Code reviewBalanced
Complex analysisReasoning
Critical decisionsReasoning

Consider Your Budget

  • High volume → Fast tier saves money
  • Quality critical → Reasoning worth the cost
  • General use → Balanced is best value

Consider Speed

  • Need results fast → Fast tier
  • Can wait for quality → Reasoning tier
  • Balance needed → Balanced tier

Model Comparison

OpenAI Models

ModelTierStrengths
GPT-4o MiniFastSpeed, efficiency
GPT-4oBalancedVersatility, quality
o1-previewReasoningDeep reasoning, accuracy

Anthropic Models

ModelTierStrengths
Claude 3 HaikuFastSpeed, safety
Claude 3.5 SonnetBalancedWriting, analysis
Claude 3 OpusReasoningComplex tasks

Meta Models

ModelTierStrengths
Llama 3.1 8BFastEfficiency
Llama 3.1 70BBalancedOpen source quality
Llama 3.1 405BReasoningFrontier open source

For Creators

When building prompt glyphs:

Default Model

Choose a sensible default:

  • Simple prompts → Fast as default
  • Most prompts → Balanced as default
  • Complex prompts → Reasoning as default

Allowed Models

Select which tiers to allow:

  • Enable all three for flexibility
  • Restrict if prompt needs specific tier
  • Consider your target audience
Last updated on: