Cufora AI

Claude, GPT & Gemini API

The original Claude, GPT and Gemini models. Far bigger token limits, a far smaller bill.

250M tokens from $10 a month instead of hitting a limit after a few hours. Requests go straight to Anthropic, OpenAI and Google over a low-latency connection. OpenAI and Anthropic compatible, works with Claude Code.

app.ts

import OpenAI from "openai"const client = new OpenAI({  baseURL: "https://www.cufora.net/api/v1",  apiKey: process.env.CUFORA_API_KEY,})const res = await client.chat.completions.create({  model: "claude-sonnet-5",  messages: [{ role: "user", content: "Hello!" }],})
claude-sonnet-5182 msgpt-5.5214 msgemini-3.1-pro196 ms
  • Anthropic Claude
  • OpenAI GPT
  • Google Gemini
  • xAI Grok
  • DeepSeek

Features

Everything you need to ship AI features. Four building blocks, one API.

Every model behind one endpoint. Call Claude, GPT and Gemini with the OpenAI SDK, the Anthropic SDK or Claude Code, and switch models with one string.

Playground

Streaming

Summarise this pull request in two sentences.

claude-sonnet-5

Adds per-key spend limits and a usage export. Existing keys keep working without a limit until you set one.

Same question, but ask gemini-3.5-flash.

gemini-3.5-flash

Ask any model…

model: auto

See every token you spend. Requests, latency and cost per key and per model, updated as they happen.

Usage

Live

Requests / min

2,940

+1.8%

Tokens today

18.4M

+4.2%

Spend today

$312

−3.1%

Spend by provider · 30 days

AnthropicOpenAIGoogle
Oct 14Oct 28Nov 12
Anthropic
OpenAI
Google

Stay up when a provider goes down. Set a primary model and a fallback, and Cufora reroutes failed calls automatically.

Routing

Active

Request

POST /chat/completions

Router

model: auto

Primary

claude-sonnet-5

92% of calls

Fallback

gpt-5.5

8% of calls

Response

streamed

Original models, not look-alikes. Every request goes straight to Anthropic, OpenAI or Google. No swapped or distilled models, new releases added as they launch.

Providers

Connected
  • Claude Opus 5 Thinking
  • GPT 5.6 Solean
  • Gemini 4 Carbon
  • Grok 4.7
  • DeepSeek V4.1 Flash
  • Composer 2.5
  • Claude Fable 5 Thinking
  • GPT 5.6 Luna
  • Gemini 4 Argon
  • Grok 4.6
  • Claude Sonnet 5 Thinking
  • GPT 5.6 Terra
  • Gemini 3.8 Flash
  • Grok 4.5
  • Claude Opus 5.5
  • GPT 5.5
  • Gemini 3.1 Pro
  • Claude Opus 5
  • GPT 5.5 Extra
  • Gemini 3 Flash
  • Claude Opus 4.8
  • GPT 5.4
  • Gemini 3.5 Flash
  • Claude Opus 4.8 Thinking
  • GPT 5.4 Mini
  • Claude Fable 5.2
  • GPT 5.4 Nano
  • Claude Fable 5.1
  • GPT 5.3 Codex
  • Claude Fable 5.1 Thinking
  • GPT 5.2
  • Claude Fable 5
  • GPT 5.1
  • Claude Sonnet 5.5
  • GPT 5 Mini
  • Claude Sonnet 5
  • Claude 4.6 Sonnet
  • Claude 4.6 Sonnet Medium Thinking
  • Claude Opus 4.7
  • Claude Opus 4.7 Thinking
  • Claude Opus 4.6
  • Claude Opus 4.6 High Thinking
  • Claude Opus 4.6 Max Thinking
  • Claude Opus 4.5
  • Claude Opus 4.5 High Thinking
  • Claude 4.5 Sonnet
  • Claude 4.5 Sonnet Thinking
  • Claude Haiku 5.5
  • Claude Haiku 5.5 Thinking
  • DeepSeek V4.1 Flash
  • Composer 2.5
  • Claude Fable 5 Thinking
  • GPT 5.6 Luna
  • Gemini 4 Argon
  • Grok 4.6
  • Claude Sonnet 5 Thinking
  • GPT 5.6 Terra
  • Gemini 3.8 Flash
  • Grok 4.5
  • Claude Opus 5.5
  • GPT 5.5
  • Gemini 3.1 Pro
  • Claude Opus 5
  • GPT 5.5 Extra
  • Gemini 3 Flash
  • Claude Opus 4.8
  • GPT 5.4
  • Gemini 3.5 Flash
  • Claude Opus 4.8 Thinking
  • GPT 5.4 Mini
  • Claude Fable 5.2
  • GPT 5.4 Nano
  • Claude Fable 5.1
  • GPT 5.3 Codex
  • Claude Fable 5.1 Thinking
  • GPT 5.2
  • Claude Fable 5
  • GPT 5.1
  • Claude Sonnet 5.5
  • GPT 5 Mini
  • Claude Sonnet 5
  • Claude 4.6 Sonnet
  • Claude 4.6 Sonnet Medium Thinking
  • Claude Opus 4.7
  • Claude Opus 4.7 Thinking
  • Claude Opus 4.6
  • Claude Opus 4.6 High Thinking
  • Claude Opus 4.6 Max Thinking
  • Claude Opus 4.5
  • Claude Opus 4.5 High Thinking
  • Claude 4.5 Sonnet
  • Claude 4.5 Sonnet Thinking
  • Claude Haiku 5.5
  • Claude Haiku 5.5 Thinking
  • Claude Opus 5 Thinking
  • GPT 5.6 Solean
  • Gemini 4 Carbon
  • Grok 4.7
  • Gemini 4 Argon
  • Grok 4.6
  • Claude Sonnet 5 Thinking
  • GPT 5.6 Terra
  • Gemini 3.8 Flash
  • Grok 4.5
  • Claude Opus 5.5
  • GPT 5.5
  • Gemini 3.1 Pro
  • Claude Opus 5
  • GPT 5.5 Extra
  • Gemini 3 Flash
  • Claude Opus 4.8
  • GPT 5.4
  • Gemini 3.5 Flash
  • Claude Opus 4.8 Thinking
  • GPT 5.4 Mini
  • Claude Fable 5.2
  • GPT 5.4 Nano
  • Claude Fable 5.1
  • GPT 5.3 Codex
  • Claude Fable 5.1 Thinking
  • GPT 5.2
  • Claude Fable 5
  • GPT 5.1
  • Claude Sonnet 5.5
  • GPT 5 Mini
  • Claude Sonnet 5
  • Claude 4.6 Sonnet
  • Claude 4.6 Sonnet Medium Thinking
  • Claude Opus 4.7
  • Claude Opus 4.7 Thinking
  • Claude Opus 4.6
  • Claude Opus 4.6 High Thinking
  • Claude Opus 4.6 Max Thinking
  • Claude Opus 4.5
  • Claude Opus 4.5 High Thinking
  • Claude 4.5 Sonnet
  • Claude 4.5 Sonnet Thinking
  • Claude Haiku 5.5
  • Claude Haiku 5.5 Thinking
  • Claude Opus 5 Thinking
  • GPT 5.6 Solean
  • Gemini 4 Carbon
  • Grok 4.7
  • DeepSeek V4.1 Flash
  • Composer 2.5
  • Claude Fable 5 Thinking
  • GPT 5.6 Luna

how it works

Create a key, call any model, and see every token you spend

  1. Step 1: Create an API key

    Sign up, add credits and create a key with its own spend limit.

  2. Write release notes for v2.4

    claude-sonnet-5

    Step 2: Call any model

    Use the OpenAI SDK you already have. Switch between Claude, GPT and Gemini with one string.

  3. Today: $0.023 across 3 models

    claude-sonnet-51,240 tok$0.011
    gpt-5.5980 tok$0.008
    gemini-3.5-flash2,310 tok$0.004

    Step 3: Track every token

    Usage, latency and cost per key and per model, live in your dashboard.

Pricing

More Claude tokens for less money

  • Starter

    For developers prototyping with frontier models

    $11.9/ month
    incl. 19% VAT
    Included250M tokens / month
    • All Claude, GPT & Gemini models
    • 2 API keys
    • 60 requests / minute
    • 7-day usage history
    • Card & crypto payments
    • Discord & Telegram support
  • Plus

    For side projects and small apps in production

    $23.8/ month
    incl. 19% VAT
    Included2B tokens / month
    • All Claude, GPT & Gemini models
    • 10 API keys
    • 300 requests / minute
    • 30-day usage analytics
    • Spend limits per key
    • Email, Discord & Telegram support
  • Pro

    Popular

    For teams shipping AI features into production

    $53.55/ month
    incl. 19% VAT
    Included10B tokens / month
    • All Claude, GPT & Gemini models
    • Unlimited API keys
    • 1,000 requests / minute
    • 90-day analytics & exports
    • Automatic provider failover
    • Priority support on every channel
  • Max

    For products with heavy and steady API traffic

    $83.3/ month
    incl. 19% VAT
    Included50B tokens / month
    • All Claude, GPT & Gemini models
    • Unlimited API keys
    • 5,000 requests / minute
    • 1-year usage history
    • Dedicated throughput
    • Dedicated account manager

Have questions?
Find answers.

  • Cufora AI is one API for Claude, GPT and Gemini. You get a single key, a single bill and one OpenAI-compatible endpoint, and you choose the model per request. Usage, cost and latency for every key and model show up in your dashboard.

  • 49 models on every plan: Claude Opus 5.5, Sonnet 5.5 and Fable 5.2 from Anthropic, including thinking variants, GPT 5.6, GPT 5.5 and GPT 5.3 Codex from OpenAI, and Gemini 4 Carbon, Gemini 4 Argon and Gemini 3.8 Flash from Google. New models are added as the providers release them.

  • No. Cufora AI speaks the OpenAI API, so any OpenAI SDK works. Set the base URL to the Cufora endpoint, use your Cufora key, and pass the model you want, for example claude-sonnet-5 or gemini-3.1-pro.

  • Every plan includes a token allowance. Monthly: Starter $10 for 250M tokens, Plus $20 for 2B, Pro $45 for 10B and Max $70 for 50B. Every plan can also be billed weekly. You can pay by card or with crypto.

  • On Pro and Max you can set a fallback model. If a call to the primary model fails, Cufora AI retries it on the fallback, so your app keeps answering.

  • Yes. From Plus upwards every API key can have its own spend limit, and usage is tracked per key and per model. That makes it easy to give each project or customer a separate budget.