NewGPT-6.2 Astra, GPT Image 2.6, Kling V4 and Kimi K3.1 Pro are now available.

Tchavi
ModelsStudioPricingDocsBlog
Tchavi
  • Why Tchavi
  • Models
  • Pricing

Tchavi is an OpenAI-compatible AI API: 100+ models, from text to video, topped up with mobile money in FCFA or your local currency, no bank card needed.

  • support@tchavi.com
  • Support

Popular models

  • GPT-6 Sol API
  • Claude Sonnet 5.5 API
  • Claude Opus 5.5 API
  • Gemini 4 Argon API
  • DeepSeek V4.1 Flash API
  • Nano Banana 2.1 (images)
  • GPT Image 2 (images)
  • Wan 3.0 (video)
  • All 100+ models

Developers

  • Quickstart
  • API Reference
  • Claude Code
  • Cursor
  • SDKs

Resources

  • Documentation
  • Blog
  • Image models
  • Video models
  • Music models
  • Audio models
  • Mobile money payments
  • Pay for AI APIs with M-Pesa
  • Claude Code with mobile money

Legal

  • Privacy
  • Terms of Service

APIs by creator

  • OpenAI API
  • Gemini API
  • Claude API
  • Qwen API
  • DeepSeek API
  • Seedance API
  • MiniMax API
  • Mistral API

Mobile money payments · Available in 12 countries: Benin, Cameroon, Congo, Côte d'Ivoire, DR Congo, Gabon, Kenya, Rwanda, Senegal, Sierra Leone, Uganda, and Zambia. Operators: Moov Money, MTN MoMo, Airtel Money, Orange Money, M-Pesa, Free Money, and Zamtel Kwacha.

© 2026 Tchavi. All rights reserved.

We use cookies to enhance your experience, provide personalised content, and analyse our traffic; see our Cookie Policy.

Back to models

Gemini 4 Carbon

google/gemini-4-carbon

Google's top reasoning model, and most of its work is invisible

Gemini 4 Carbon is Google's top reasoning tier, released on 10 November 2026. Google built it for complex software engineering, maths and science, and multi-step agents. It takes text, images, audio, video and PDFs, returns text, and reads up to 1,048,576 tokens of context.

Modalities
Input / Output
8.8 / 44cr / 1K
Context window
1M
Released
Nov 10, 2026
Capabilities
  • Tools
  • JSON mode
  • Vision
  • PDF
  • Streaming

Last updated Nov 19, 2026

It always thinks before it answers, and that thinking cannot be switched off. The thinking is billed as output tokens and you do not see it. In our own test, an exact counting puzzle came back right after 7,905 billed output tokens, of which about 7,723 were thinking: 97.7% of what was billed never appeared on screen, for a visible answer of 182 tokens. So the output price per million tokens tells you little about what a request will cost. Test it on your own prompts before you plan a budget.

On Tchavi you pay with mobile money, no bank card and no KYC. One account and one API key open 100+ models, and your credits never expire.

What this model does well

  • 🧠

    Always-on reasoning

    It thinks through hard problems before answering, and you cannot turn that off

  • 📚

    1M-token context

    A codebase or a stack of long documents in a single call

  • 🎞️

    Text, image, audio, video, PDF

    Mixed input in one request, text out

  • 🔧

    Tools and JSON mode

    Function calling and structured output for your agents

Good to know

  • Most of the output you pay for is thinking you never see. In our test, 97.7% of the billed output tokens were reasoning (about 7,723 of 7,905), against a 182-token visible answer. The cost of a request depends on how long the model thinks, not on how long the answer is, so you cannot predict it from your prompt or from the reply. Measure a few real requests first.
  • The thinking cannot be switched off, so it is not the pick for quick, cheap replies. Our counting test took 93.6 seconds.
  • Answers are capped at 32,768 output tokens on Tchavi, below the 65,536 Google documents, so a request can never ask for more than the endpoint serving it accepts. Leave room for the hidden reasoning when you set max_tokens.
We never save what you send to a model or what it returns. We log only your settings (like size or voice) and how much was used, for billing. Files you attach in the playground are deleted after 24 hours.

Technical specifications

Provider
google
API identifier
gemini-4-carbon
Category
Chat & Text
Context window
1,048,576 tokens
Max output
32,768 tokens
Billing unit
token
Streaming
Yes

Ideal for

  • Complex software engineering
  • Maths and science problems
  • Agents that chain many steps
  • Counting, planning and constraint puzzles
  • Reading long documents, images, audio and video together

Ready to try it?

Test this model in the playground, no setup needed.

Frequently asked questions

Why is the bill bigger than the answer I see?

Gemini 4 Carbon always reasons before it answers, and that reasoning is billed as output tokens even though it is not shown. In our test, 7,723 of 7,905 billed output tokens (97.7%) were thinking, for an answer of 182 tokens. The more demanding the problem, the more it thinks.

Can I turn the thinking off?

No. Reasoning is always on for this model. If you want short, fast replies, pick a lighter model from the catalogue.

How much context and output does it have?

Up to 1,048,576 tokens in. Output is capped at 32,768 tokens per request on Tchavi, thinking included in what is billed, so leave headroom for it.

What can I send it?

Text, images, audio, video and PDFs, and it returns text.

How do I call Gemini 4 Carbon with my SDK?

Swap the base URL for Tchavi's, use your Tchavi key and set model: "gemini-4-carbon" on /v1/chat/completions.

Can I pay with mobile money, without a bank card?

Yes. Top up your Tchavi account with Orange Money, MTN MoMo, M-Pesa or another mobile money operator. No bank card is needed. Credits land as soon as the payment is confirmed and are deducted as you use them.

Is the API compatible with OpenAI's?

Yes. Keep your current OpenAI SDK, just change the base URL to Tchavi's (shown in the API tab) and use your Tchavi key. The rest of your code stays the same.

API

import Tchavi from '@tchavi/sdk';

const client = new Tchavi({ apiKey: 'API_KEY' });

const response = await client.chat.completions.create({
  model: 'gemini-4-carbon',
  messages: [
  {
    "role": "user",
    "content": "Hello!"
  }
],
});

console.log(response.choices[0].message.content);
console.log('Credits used:', response.tchavi.credits_used);

API parameters

Every parameter this model accepts on the API. Check the SDK for full typings.

modelRequired
string

ID of the chat model (e.g. `gemini-4-carbon`).

messagesRequired
Array<{ role, content }>

Conversation messages. Each item has a `role` (system/user/assistant) and a `content` that's either a plain string or an array of content blocks (`text`, `image_url`, `document_url`).

temperature
number
Default: 1

Sampling temperature between 0 and 2. Higher values = more random.

max_tokens
integer

Maximum number of tokens to generate in the response.

max_completion_tokens
integer

OpenAI alias for `max_tokens`. Either field works; if both are sent, `max_completion_tokens` wins.

stream
boolean
Default: false

If true, partial message deltas are streamed as Server-Sent Events.

tools
Array<{ type, function }>

Function-calling toolbox. Pass an array of `{ type: "function", function: { name, description, parameters } }`. Returned tool calls land in `choices[].message.tool_calls`. Only models with `supportsTools: true` accept this.

tool_choice
"auto" | "none" | { type, function }

How the model picks among tools: `"auto"` (default), `"none"` to force a plain text reply, or `{ type: "function", function: { name } }` to force a specific tool. Only with `supportsTools: true`.

response_format
{ type: "json_object" | "text" }

Constrain the output. Pass `{ type: "json_object" }` to force valid JSON (the model still needs a JSON instruction in your prompt). Only models with `supportsJsonMode: true` honour this.

Pricing

per 1,000 tokens
8.8 credits
Input 8.8 · Output 44 cr/1K

How billing works

  • Billed per token used, down to a thousandth of a credit.
  • A typical message (500 in, 150 out) costs about —, or 11 credits.
Real-world scenarios
Per day
Per week
Per month

Estimated over 30 days, for a typical message: ~500 input + ~150 output tokens, billed to the thousandth of a credit.

A 10 000 F CFA top-up gives you about ~2 727 questions. Pay with Orange Money, MTN MoMo or M-Pesa, no Visa card needed.

Ready to use Gemini 4 Carbon?

Create your accountAPI Documentation

Bonus credits on your first top-up · No card to sign up