30 chat models · 2 image models
Many models. One conversation.
Frontier reasoning, multimodal vision, and image generation in a single chat. Live model availability, streaming responses, no clutter.
Live model catalog
Pulled directly from NVIDIA NIM at request time. Each model has different strengths — pick the right one for the task.
Dbrx Instruct
open weightsDatabricks model.
Databricks·128K ctx
DeepSeek V4 Flash
284B MoE · 13B activeFast reasoning. Good default for general queries.
DeepSeek·1000K ctx
DeepSeek V4 Pro
1.6T MoE · 49B activeMaximum reasoning, slowest. Best for hard problems.
DeepSeek·1000K ctx
Gemma 4 31B It
open weightsGoogle model.
Google·128K ctx
GLM 5.1
754B MoE · 40B activeOpen-source coding & agentic engineering flagship.
Z.ai·200K ctx
GPT-OSS 120B
120B open weightsOpenAI's open-weight 120B.
Streaming responses
Every token as soon as the model produces it. No spinner-and-wait.
Live artifacts
HTML, React, SVG, Mermaid, code — rendered next to the chat in a sandboxed view.
Multimodal input
Drag in images or PDFs. Multimodal models read them natively.