Models
Vercilio routes across the world's leading conversational AI providers — OpenAI, Anthropic, Google, Meta, xAI, DeepSeek, Mistral, and more — so you can move between frontier reasoning and fast everyday work in the same window.
We follow new releases as they ship. Every model below is currently available in chat to subscribers, with appropriate guardrails for vision, file reading, and web search where supported.
The full GPT lineup — fast minis for everyday questions, balanced 4o-class models, and frontier GPT-5 for reasoning that benefits from a wider window.
Claude Haiku for quick rewrites, Claude Sonnet as a dependable default, and Claude Opus when the answer needs careful, long-form thinking.
Gemini Flash for speed, Gemini Pro for harder questions, and Gemini Pro Preview for the very latest from DeepMind — multimodal-native and search-aware.
Llama models for open-weight workflows when you want a model that runs the same way many teams already self-host.
Grok for a different point of view, with strengths in current events and casual conversational tone.
DeepSeek's frontier-quality reasoning models at noticeably lower cost — useful for long analysis sessions without watching the meter.
Mistral and Magistral covers strong European-trained models — fast, multilingual, and tuned for production code and writing tasks.
Command models built for retrieval and enterprise workflows, with a focus on multilingual document work.
Search-grounded Sonar models when the answer should come with citations from the live web.
Names change every quarter. The categories below stay the same — pick the one that matches the work, and the picker filters down.
Short answers, quick rewrites, format conversions, and everyday Q&A. Lowest per-message cost; great for high-volume work and free-plan use.
The models you'd reach for without thinking — strong quality, fair latency, sensible cost. This is where most paid traffic goes.
Multi-step problems, deeper analysis, long context. Higher per-message cost, noticeably better answers on harder questions.
The latest releases from OpenAI, Anthropic, and Google. Use when an answer needs to be as good as today's best general-purpose AI can produce.
Drop in a screenshot, photo, diagram, or whiteboard shot. Vision-capable models read images directly inside the conversation.
Attach PDFs and other documents. The model parses them inline — no separate uploader, no chunking dance.
Toggle search and the model answers with live citations. We surface the sources beneath every reply.
On reasoning-capable models you can choose Low, Medium, or High. Higher effort means longer thinking and better answers on hard tasks.
Let Vercilio pick the right model for each message. Our router uses live performance signals to send you the best fit per turn.
Compatible models can call tools when needed. Currently used for web search; agentic workflows are on the roadmap.
The free plan ships with 100 credits a month on the small-tier models — perfect for learning the product. Upgrade to a paid plan when you want frontier reasoning, vision, web search, and the monthly environmental contribution.
