Vamos gives you access to the latest models from Anthropic, OpenAI, Google, and xAI. You don't have to be an AI expert to pick the right one, just leave the model selector on AUTO and Vamos handles it.
This article covers what's available, how AUTO works under the hood, and when you'd manually override it.
Just use AUTO
AUTO is the default — and the right choice for 95% of what you do.
How AUTO decides:
Writing tasks (scripts, captions, posts, emails, anything requiring your brand voice) → Claude Sonnet 4.6. Claude has the best voice / nuance of any model.
Research tasks (pulling videos, analyzing comments, scraping profiles, fetching URLs, summarizing large content) → better agentic models like GPT-5.4.
Simple questions (quick factual asks, short explanations) → smaller models.
Complex reasoning (deep strategy work, multi-step analysis) → larger models with thinking mode auto-enabled.
The result: you get Claude-quality writing when it matters, and you don't burn Claude credits on a task cheaper models could do in a third of the cost. The output quality is still optimized, Vamos routes to whichever model is actually best for the sub-task, not just the cheapest available.
You don't need to think about thinking mode. Vamos enables extended reasoning automatically when the task is complex enough to benefit from it.
The full model lineup
Anthropic (Claude)
Claude Sonnet 4.6 — Top-tier writing quality, fast. AUTO uses this for most writing tasks.
Claude Opus 4.6 — The smartest model available. Best for complex reasoning, strategy, or writing where you want absolute maximum quality. Highest credit cost.
OpenAI (GPT)
GPT-5.4 — Strong all-rounder. Good reasoning, good writing, good research.
GPT-5.4 mini — Faster and cheaper version. Great for bulk research and content analysis where raw speed matters.
Google (Gemini)
Gemini 3 Flash — Fastest, cheapest model. Great for research, summarization, and simple tasks.
Gemini 3.1 Pro — Larger context handling (great for long videos, long docs, massive competitor research runs).
xAI (Grok)
Grok 4 — Useful when you want a different angle or more "edge" in creative work.
Quick comparison
Model | Best for | Speed | Credit cost |
Claude Sonnet 4.6 | Writing (scripts, captions, emails) | Fast | High |
Claude Opus 4.6 | Hardest reasoning / best writing | Slower | Highest |
GPT-5.4 | Balanced general use | Fast | Mid |
GPT-5.4 mini | Bulk research, analysis | Very fast | Low |
Gemini 3 Flash | Simple tasks, fast research | Very fast | Lowest |
Gemini 3.1 Pro | Long content / large context | Fast | Mid |
Grok 4 | Creative edge / unique voice | Fast | High |
Context windows
Context is how much text (chat history, uploaded files, linked content) a model can "see" at once. Vamos uses each model's published context window:
Claude Sonnet 4.6: 1M tokens (~700,000 words)
Claude Opus 4.6: 1M tokens
GPT-5.4: 400K tokens
GPT-5.4 mini: 400K tokens
Gemini 3 Flash: 1M tokens
Gemini 3.1 Pro: 2M tokens
Grok 4: 256K tokens
For long chats, pasted YouTube channels, big file uploads, or massive competitor runs, Gemini 3.1 Pro gives you the most room. AUTO will route there when it detects the input is large enough to need it.
When to override AUTO
Most of the time you shouldn't. But there are a few cases:
You want Claude Opus specifically — for the highest-stakes writing work or complex strategy reasoning. Opus costs more but for flagship pieces it's worth it.
You want a specific voice angle — Grok 4 sometimes produces outputs with a different flavor than Claude. Worth trying if AUTO output feels too "polished."
You're low on credits — force Gemini 3 Flash or GPT-5.4 mini to stretch your balance.
To switch, click the model selector in the chat input and pick manually. Vamos remembers your selection for the current chat.
Saving credits
The biggest lever is just leaving AUTO on. Two other tips:
Let Vamos research with a cheap model, then write with Claude. If you prompt "Research X, then write me scripts," AUTO will automatically use a cheap model for the research phase and switch to Claude for the writing phase inside the same response.
Be specific. Clear prompts finish in one pass. Vague prompts trigger clarification back-and-forth that burns credits.
See Plans, pricing & credits explained for the full credit breakdown.
Common issues
Output feels generic / not in your voice — Vamos is almost always using Claude for writing. If it feels off, check your Brand Voice is set up — see Brand Profile: Getting AI to sound like you. You can also tell Vamos "I hate this style" and it'll save the correction to Memory.
You want a specific model's strengths — manually select it. AUTO is a great default, not a cage.
Questions? Message us in the chat widget or email contact@getvamos.ai.