LLM Component Database
This page serves as a component registry for Large Language Models. When designing your application “builds,” use this data to select models based on context requirements, cost limits, and search capabilities.
Technical Specifications Table
Section titled “Technical Specifications Table”| Model Name | Provider | Context Window | Input Cost (per 1M) | Output Cost (per 1M) | Core Strengths (Stat Modifiers) |
|---|---|---|---|---|---|
| Gemini 2.5 Pro | 2,000,000 | $1.25 | $5.00 | Multi-modal reasoning, large file ingestion, long context RAG | |
| Gemini 2.5 Flash | 1,000,000 | $0.075 | $0.30 | Speed, high rate-limits, lightweight agent tasks, cost efficiency | |
| Claude 3.5 Sonnet | Anthropic | 200,000 | $3.00 | $15.00 | Code generation, complex agent logic, structured JSON output |
| GPT-4o | OpenAI | 128,000 | $2.50 | $10.00 | High general reasoning, conversational speech, fast API response |
| Llama 3.3 70B | Meta (Open) | 128,000 | Self-Hosted | Self-Hosted | Local privacy, fine-tuning potential, high reasoning output |
Active Passive Modifiers (Features)
Section titled “Active Passive Modifiers (Features)”When selecting a model, pay attention to the passive capabilities they unlock:
🚀 Google Search Grounding
Section titled “🚀 Google Search Grounding”- Unlocked by: Gemini series.
- Effect: Allows the model to search the web in real-time to verify facts, reducing hallucinations to near-zero for factual queries.
💻 Computer Use / Tool Calling
Section titled “💻 Computer Use / Tool Calling”- Unlocked by: Claude 3.5 Sonnet, GPT-4o.
- Effect: High accuracy in parsing tool schemas and calling external APIs sequentially in agent loops.
🧪 System Instruction Guarding
Section titled “🧪 System Instruction Guarding”- Unlocked by: Most frontier models.
- Effect: Strong adherence to system prompts and constraints under adversarial user inputs.