LLM
Large language models exposed through the same API as our image and video models — useful as pipeline glue, structured data extraction, or design-time Q&A about a generated output. Today: Google's Gemini 2.5 Flash as the multimodal flagship, plus Llama 3.2 Vision and Qwen3-VL via Ollama for open-source local-friendly stacks.
smart_toy
psychologyLLM
google/gemini-3.7-flash
Gemini 3.7 Flash is Google's workhorse multimodal LLM: send a prompt plus up to 11 images and get back copy, structured JSON, captions or analysis. Its sharpest edge over earlier Flash models is how closely it follows a brief — visual references, brand rules and dense documents — with a reasoning dial you set per call, from quick low-effort answers to careful multi-step work.
Google
bolt1~10s
Try it