| Company | Basic / General-Purpose Models | Reasoning / Agentic Models |
|---|---|---|
| Open AI | GPT-5.2 or GPT 5.3 Instant | GPT-5.3 (thinking / “GPT-5 Pro” deep-reasoning variant) |
| Anthropic | Claude Haiku 4.5 (fast / cost-efficient general use) | Claude Sonnet 4.5 ((intermediate) complex reasoning, agents / coding) Claude Opus 4.5 ((advanced) complex reasoning, coding, agentic & enterprise tasks) |
| Gemini 2.5 (fast / literal general-purpose / multimodal) | Gemini 3.1 Pro (and Deep Think) (reasoning / strategy / vibe coding) | |
| Meta | Llama 4 Scout (lightweight / general-purpose) | Llama 4 Maverick (stronger reasoning / coding / larger-scale agentic potential) |