Your models. Your tools.
One place to get started.

Connect with three values: a base URL, your API key, and an enabled model identifier. Vani exposes bounded Chat Completions, stateless Responses and Anthropic Messages adapters at the endpoint below.

For pay-as-you-go access, open the API platformto manage prepaid keys, check your balance, and see API usage by model and key for today, 7, 30 or 90 days. Eligible API models cost 30% below their verified official rates; private models priced directly by Vani are labelled “Vani price”. Rates include source references and context tiers; unavailable rates are never treated as free.

export VANI_BASE_URL="https://api.vani.ai/v1"
export VANI_MODEL="openai/gpt-6-sol"
# Set VANI_API_KEY in your shell secret store or environment.
# Use the API origin shown here, ending in /v1 once.
# The website /chat and /api/chat routes are not this API base.
# Start by fetching /models; that check does not invoke inference.

curl --fail-with-body "$VANI_BASE_URL/models" \
  -H "Authorization: Bearer $VANI_API_KEY"

# Use an exact data[].id from this account-scoped response.

curl --fail-with-body "$VANI_BASE_URL/chat/completions" \
  -H "Authorization: Bearer $VANI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/gpt-6-sol","messages":[{"role":"user","content":"Hello"}],"max_completion_tokens":256,"stream":false}'

A curated set of 15 popular clients and coding tools. Each guide separates supported API settings from features that need an adapter or extra validation.

Coding agents

Hermes Agent

Configure a named custom provider using Hermes' documented Chat Completions transport.

Chat Completions setup

View setup

Coding agents

Codex

Use the Codex CLI with Vani’s stateless Responses adapter, including local function and namespaced custom tools.

Responses adapter

View setup

Coding agents

OpenCode

Add Vani as a custom provider with the OpenAI-compatible Chat Completions package.

Chat Completions setup

View setup

Coding agents

Claude Code

Connect Claude Code to Vani’s Anthropic Messages adapter for text, streaming and client-executed tools.

Messages adapter

View setup

Editors

Cursor

Cursor's current official BYOK guide documents provider keys, but does not verify this custom Vani endpoint or GPT-6 Sol.

Check client version

View setup

Editors

Cline

Use Cline's OpenAI Compatible provider and the exact Vani model identifier.

Chat Completions setup

View setup

Editors

Continue

Configure a Chat-only OpenAI-compatible model and explicitly disable automatic Responses routing.

Chat Completions setup

View setup

Coding agents

Aider

Point Aider's OpenAI-compatible transport at Vani and preserve the exact Vani model ID.

Chat Completions setup

View setup

Editors

Zed

A fast code editor with its own agent and configurable OpenAI-compatible model providers.

Custom provider · validate tools

View setup

Coding agents

OpenClaw

Connect the OpenClaw agent runtime to Vani through a custom Chat Completions provider.

Text chat setup · configuration checked

View setup

Chat workspaces

Open WebUI

Use an admin-managed OpenAI-compatible connection with the Vani model allowlist.

Text chat setup · configuration checked

View setup

Chat workspaces

LibreChat

Add Vani as a custom OpenAI-compatible endpoint using a minimal request baseline.

Text chat setup · configuration checked

View setup

Chat workspaces

LobeHub / LobeChat

Configure the OpenAI-compatible provider and an exact custom Vani model id.

Text chat setup · configuration checked

View setup

Chat workspaces

Cherry Studio

Use a custom OpenAI endpoint for ordinary chat; preserve the Vani model prefix.

Text chat setup · configuration checked

View setup

Chat workspaces

AnythingLLM

Choose the Generic OpenAI LLM provider for text chat while keeping document indexing separate.

Text chat setup · configuration checked

View setup

1. Open New chat, choose an available model, and send your message. You can change models as you continue.
2. Use the paperclip to attach a file. Turn on Image or Videounder the message box to create media from your words; choose the model (and a video's length) in the bar that appears. The line under the message box always says what Send will do and what it uses from your allowance.
3. Open Tool settings to choose whether Vani creates requested images and videos right away or asks first, how prompts are handled (Enhance prompt lets your chat model add visual details; Use my words sends your message unchanged), and your preferred models. Creating right away uses your allowance, up to one media generation per reply. Web tool proposals always ask first. Completed images and videos are ready to use without a continuation step.
4. Use Stop to request cancellation. Work already completed can still count toward usage.

Media library is an archive of generated results, while creation stays in chat.

Source links and citations help you check the evidence. Search results and web pages are untrusted reference material, and generated summaries can be incomplete or wrong. Open the cited source before relying on a claim.

Create a Project for related conversations and reusable instructions. Changes apply to future sends; they do not rewrite saved messages or a request already being processed.

• Archive a project to stop new assignments and new project chats. Existing conversations keep the project instructions.
• Delete a project to detach its conversations. The saved chats remain available.
• If another tab updates the project first, reload its latest version while preserving your unsaved edits, review them, then save again.
• Use Edit, Regenerate or Branch on a saved message to create a separate conversation. The original stays intact; edited or regenerated responses preserve a saved branch draft. If a draft is waiting, use Generate response to start it.

Resolve active work before branching. A missing or expired original attachment can block a draft. After an uncertain send, refresh saved results instead of repeatedly generating the same draft.

Your subscription allowance is shared across available models and conversations. The app applies both billing-period and weekly limits, along with request-rate and concurrency limits. A higher plan changes configured capacity; it does not guarantee a fixed number of messages, tokens or images.

Your usage dashboard offers Today, 7 days and 30 days as UTC calendar windows including today. Request counts use their start date; token and allowance records use settlement time, so activity spanning midnight can appear on different days.

• Provider-reported tokens exclude local estimates. Records without provider reports are labelled separately.
• Active work can reserve allowance. Reset times are scheduled dates, not a prediction of when capacity will run out.
• The model-rate calculator explains subscription allowance debits per million input/output tokens. On Claude Sonnet 5.5, Claude Fable 5.1, MiMo V2.6 Flash, MiMo V2.6 Pro, Kimi K2.8 Preview and Claude Opus 5.5, the part of a prompt the provider serves from its cache uses a fraction of the usual allowance; on other models cached input counts at the full input rate.
• API keys created in Vani Chat use this same allowance; API platform keys use a separate prepaid balance. Chat allowance values are not cash balances or provider invoices. Automatic API top-ups require configured payment processing; check the API platform for funding availability.

Configure a generic OpenAI-compatible client

1. Select the protocol supported by your client guide: Chat Completions, Responses, or Anthropic Messages.
2. For OpenAI-compatible clients, use the base URL ending in /v1. Anthropic SDKs that append /v1/messages use the origin without /v1 instead. Follow your client guide, then set your API key and an enabled model identifier.
3. Disable unsupported extras such as raw media input, response_format, extended thinking, stored Responses state and automatic provider fallbacks.
4. Start with a short text request, then enable streaming, then test the client’s own function tools.

Chat Completions accepts function definitions, but your API client executes them. Vani does not automatically run arbitrary client functions. Validate the model’s proposed arguments, apply your own permissions, execute the function, then send the assistant tool call and a matching tool result in the next request.

{
  "model": "openai/gpt-6-sol",
  "messages": [
    {
      "role": "user",
      "content": "What is the weather in Paris?"
    }
  ],
  "max_completion_tokens": 256,
  "tools": [
    {
      "type": "function",
      "function": {
        "name": "get_weather",
        "description": "Get weather for a city",
        "parameters": {
          "type": "object",
          "properties": {
            "city": {
              "type": "string"
            }
          },
          "required": [
            "city"
          ],
          "additionalProperties": false
        }
      }
    }
  ],
  "tool_choice": "auto"
}

Every assistant tool call must have a matching result before another message. Built-in web research and media actions in the Vani chat UI are a separate workflow with saved action state. In Tool settings, Vani can create requested images and videos right away or ask first. Web tool proposals always require confirmation.

Vani Chat can show charts, diagrams, formulas and clickable choices in a reply. Each is a plain-text block inside the assistant message, so the API returns it exactly as written, and nothing is added to API requests. To get these blocks from the API, put the instruction below in your system message, then render the blocks in your client, or show them as code.

Rich replies: this chat draws the fenced blocks below. Use one only when it clearly helps
(comparing numbers, a trend, a process, a formula, a choice the user must make); most replies
need none. Use the exact language tag, never nest a block inside another fence, and explain in
prose around it.

Math: LaTeX inline as $…$ and display equations as $$…$$ on their own lines; in a reply with
math, write a literal dollar sign as \$.

Charts: a vani-chart block holding one JSON object, e.g. {"type":"bar","title":"Revenue by
quarter","x":{"label":"Quarter","values":["Q1","Q2","Q3"]},"y":{"label":"Revenue","unit":"$"},"series":[{"name":"2025","values":[120,135,160]}]}.
type is line, bar, area, pie or scatter; each series has one number or null per x value;
optional "description", "source", "stacked":true (bar, area), "horizontal":true (bar). pie:
x.values are the slice labels and there is exactly one series. scatter: no x.values; each series
has "points":[[x,y],…]. Up to 8 series and 500 values each. Chart only real numbers from the
conversation or the user's files, never invented ones; aggregate large data and say so in
"description".

Diagrams: a mermaid block using only flowchart (TD or LR), sequenceDiagram, stateDiagram-v2,
mindmap or pie, with up to 40 nodes and short labels; no styling, classDef, click or HTML.

Choices: when you need the user to pick between concrete options before you can continue, end
the reply with one vani-input block holding one JSON object, e.g.
{"type":"single","question":"Which plan should I
detail?","options":["Basic","Pro","Enterprise"]}. type is single (one click), multiple
(checkboxes; optional "min" and "max"), rank (the user orders the options) or form with 2 to 5
"fields":[{"name":"budget","label":"Budget","type":"number","required":true}] (field types text,
textarea, number, select with "options", date, email). Give 2 to 8 options with short labels; an
option may be {"label":"…","description":"…"}. The answer arrives as the user's next message
(e.g. "Chose: Pro") and the user may type something else instead. At most one vani-input per
reply, never to confirm a tool action.

Charts and diagrams show data and structure only: they never replace a requested picture or
video, which only the image and video tools may make, and neither do SVG or drawing code.

```vani-chart
{
  "type": "bar",
  "title": "Revenue by quarter",
  "description": "Quarterly revenue, 2025 vs 2026.",
  "x": {
    "label": "Quarter",
    "values": ["Q1", "Q2", "Q3", "Q4"]
  },
  "y": {
    "label": "Revenue",
    "unit": "$"
  },
  "series": [
    {
      "name": "2025",
      "values": [120, 135, 160, 172]
    },
    {
      "name": "2026",
      "values": [140, 151, 189, null]
    }
  ],
  "source": "Finance export, October 2026"
}
```

Closing the compatibility stream requests cancellation; already produced output can still be settled. Read error frames even after HTTP 200. The status pageseparates recent model outcomes from low-sample or unknown observations. It shows Vani's recorded attempts, not official global provider uptime, and does not invent recovery estimates.

Saved chats, projects, uploaded files, research evidence and in-app updates use this deployment’s database and storage. Selected model providers and web sources may still receive the request data needed for inference or search. Self-hosted application storage does not mean inference runs on this server.

Temporary chat is not added to saved conversation history, but operational run, usage and security records can still exist. Uploaded files have their own retention lifecycle. This is not a zero-retention promise.

Use Report a bug in the sidebar, or Report this error beside a failed message. The report contains the text you enter, an optional page path and an optional request reference. Conversations and files are not attached automatically.

A request reference helps staff correlate operational metadata; it is not proof of request ownership. Keep secrets and sensitive content out of bug reports. Manage keys in Developer settings and review the privacy policy.