Skip to content
Vichar
26 models, one OpenAI-compatible API

Build with every AI model through one API

Frontier models from 11 labs behind a single key, with automatic failover, per-key spend limits and real-time cost analytics. Pay the listed per-token rate.

Get your API keyRead the docs
  • No subscription
  • 0% token markup
  • Failed calls are free
Playground

Explain an API gateway in one sentence.

Latency

…

Tokens

…

Cost

…

Sample run · cost uses this model's live per-token price

Frontier models from the labs you already use

OpenAI
Google
DeepSeek
Moonshot
Z.ai
Alibaba
Meta
Mistral
NVIDIA
AI Singapore
IBM

Everything you need to ship AI to production

One integration gives you the whole frontier, with the controls a production workload needs.

One API for every model

Switch labs by changing one string. Same OpenAI-compatible request, same SDK, same balance.

client.chat.completions.create({

model:"kimi-k3"

messages,

})

Failover built in

If a route errors or times out, the request is retried on another route for the same model.

Route Atimeout
Route B200 OK

Retried automatically · failed attempt not billed

Spend limits per key

Cap each key with a monthly budget and an expiry, so a runaway job can never drain your balance.

production-backend

vc_7fK…a91 · expires in 30d

$42.18 of $100 / month

The key pauses when it reaches its limit

See every token and dollar

Usage, cost and latency per key, project, model and provider, updated as requests happen.

Spend · last 12 days

$184.20

InputOutput

One request. Any model.

Your app sends one request. Vichar routes it to the lab that serves the model you asked for, with the same request and response shape every time.

26

Models

11

Labs

1.1M

Max context

0%

Markup

example · kimi-k3 → Moonshot · 200 · 712 ms
POST /v1/chat/completionsPOST /v1/messagesGET /v1/models

For developers

Change two lines.
Keep your SDK.

Streaming, tool calling and structured outputs work the same way across every lab. The Anthropic Messages format works too.

1import OpenAI from "openai";
2
3const client = new OpenAI({
4−  apiKey: process.env.OPENAI_API_KEY,
5+  baseURL: "https://api.vichar.io/v1",
6+  apiKey: process.env.VICHAR_API_KEY,
7});
8
9const res = await client.chat.completions.create({
10  model: "kimi-k3",
11  messages: [{ role: "user", content: "Hello" }],
12  stream: true,
13});

Your whole AI spend, in one dashboard

Requests, tokens, latency and cost for every key, project, model and provider, updated in real time.

Cost per key and model

Find the expensive workload in seconds.

Keys with budgets

Spend caps, expiry and one-click revoke.

Errors and reliability

Error rates and retries per provider.

Start in minutes

Step 1

Create an API key

Sign up, add credits from $20, and create a key with its own spend limit and expiry.

Step 2

Change your base URL

Point any OpenAI-compatible SDK at api.vichar.io/v1. The rest of your code stays the same.

Step 3

Ship on any model

Pick models per request, and watch tokens, latency and cost per key in the dashboard.

Frontier models, ready to call

Live per-token pricing from the catalogue.

Browse all 26 models

Input / output per 1M tokens

Simple, honest pricing

Listed model rates, a flat fee when you add credits, and nothing for requests that fail.

0%

Token markup

Every request is billed at the model's published input and output rate.

5%

Top-up fee

A flat platform fee when you add credits. A $20 top-up costs $21.

$0

For failed requests

Requests that fail before a provider answers never deduct credits.

See pricing and cost calculator

Private and secure by default

We keep the metadata needed to bill you, nothing more.

Prompts are not stored

By default we keep only billing metadata: tokens, cost, latency and provider.

Keys hashed at rest

Keys are stored as one-way SHA-256 hashes and shown in full only once.

Encrypted in transit

HTTPS only, TLS 1.2+, between your app, Vichar and every provider.

Questions, answered

  • Yes. Point your client at https://api.vichar.io/v1 and use a Vichar API key. Your existing code keeps working.

Something else? Contact us

Start building with every model today

One key, 26 models, and you only pay for what you use.

Get your API keyRead the docs