> ## Documentation Index
> Fetch the complete documentation index at: https://docs.laozhang.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Gemini 4 Carbon API: Pricing, Code Examples, and Carbon vs Argon

> Gemini 4 Carbon (gemini-4-carbon) is live on LaoZhang API with a default-group API key, at Google's price of $2 input and $10 output per 1M tokens. Model ID, the December 1 price change, curl and Python examples, thinking levels, and how it differs from Gemini 4 Argon.

Gemini 4 Carbon (`gemini-4-carbon`) is the Gemini model Google made generally available on November 10, 2026, built for hard coding, debugging and science reasoning. LaoZhang API serves it from November 11 to any `default`-group key at \$2 input and \$10 output per 1M tokens, matching Google's price and Gemini 4 Argon.

* **Published**: November 11, 2026
* **Last checked**: November 11, 2026
* **Status**: Active; available in the `default` group

## Key facts

| Item | Details |
| - | - |
| Model ID | `gemini-4-carbon` |
| Google release | November 10, 2026 (generally available) |
| Live on LaoZhang API | November 11, 2026 |
| OpenAI-compatible endpoint | `POST https://api.laozhang.ai/v1/chat/completions` |
| Gemini-native endpoint | `https://api.laozhang.ai/v1beta/models/gemini-4-carbon:generateContent` |
| Token group and billing | `default` group, usage-based (per token) |
| Current price | \$2 input, \$10 output, \$0.1 cache read per 1M tokens |
| Context window (Google) | 1,048,576 tokens |
| Max output per request (Google) | 65,536 tokens by default |
| Thinking levels | `low`, `medium`, `high`; Google's default is `medium` |
| Works with | Function calling, structured output (JSON Schema), image input, streaming |

## Pricing

Unit prices from the public pricing configuration on November 11, 2026, in US dollars per 1M tokens:

| Item | LaoZhang API now | Google, through Nov 30 | Google, from Dec 1 |
| - | -: | -: | -: |
| Input | \$2 | \$2 | \$4 |
| Output (thinking included) | \$10 | \$10 | \$20 |
| Cache read | \$0.1 | \$0.1 | \$0.2 |

* Google's introductory price runs through November 30, 2026. From December 1, Google charges \$4 input and \$20 output. LaoZhang API bills at Google's price and will move on the same date, with a notice in [Announcements](https://docs.laozhang.ai/en/changelog.md) beforehand. Gemini 4 Argon changes on the same day by the same amount.
* Thinking tokens are billed at the output price. A request with 46 input tokens, a 7-token answer and 188 thinking tokens is billed for 195 output tokens, about \$0.002.
* Cache reads come from Google's implicit caching. When you resend the same prefix of 4,096 tokens or more, the cached part is billed at the cache price automatically; no extra parameter is needed.
* Token usage and the charge for every request are in the [usage logs](https://api.laozhang.ai/log). Current prices live in [Models & Pricing](https://docs.laozhang.ai/en/models.md#vendor-google-text).

## Gemini 4 Carbon vs Gemini 4 Argon

| | `gemini-4-carbon` | `gemini-4-argon` |
| - | - | - |
| Google release | November 10, 2026 | September 30, 2026 |
| Price per 1M tokens | \$2 input, \$10 output | \$2 input, \$10 output |
| Context window | 1,048,576 tokens | 1,048,576 tokens |
| Thinking levels | `low`, `medium`, `high` | `low`, `medium`, `high` |
| Thinking tokens per turn | More (per Google) | Fewer |

* Both models share the same per-token price, context window, built-in tools and request format. Switching is a one-word change to `model`.
* Google notes that Carbon spends more thinking tokens per turn. The same request usually produces more output tokens on Carbon, so it costs more in total and takes longer.
* Reach for `gemini-4-carbon` on hard coding, debugging, math and science problems. Stay on `gemini-4-argon` for work it already handles well, and for high-volume or cost-sensitive traffic. Before switching production traffic, run your own samples through both and compare quality, latency and the token counts in the [usage logs](https://api.laozhang.ai/log).

## Code examples

Create an API key in [Token management](https://api.laozhang.ai/token): pick the `default` group and set the billing mode to Usage first (recommended) or Usage-based. Then export it in your shell:

```bash theme={null}
export LAOZHANG_API_KEY="your-laozhang-api-key"
```

> [tab] OpenAI-compatible
>
> ```bash theme={null}
> curl https://api.laozhang.ai/v1/chat/completions \
>   -H "Authorization: Bearer $LAOZHANG_API_KEY" \
>   -H "Content-Type: application/json" \
>   -d '{
>     "model": "gemini-4-carbon",
>     "reasoning_effort": "high",
>     "messages": [
>       {"role": "user", "content": "In three sentences, explain what a database index is and why it slows down writes."}
>     ]
>   }'
> ```

> [tab] Gemini native
>
> ```bash theme={null}
> curl "https://api.laozhang.ai/v1beta/models/gemini-4-carbon:generateContent" \
>   -H "Authorization: Bearer $LAOZHANG_API_KEY" \
>   -H "Content-Type: application/json" \
>   -d '{
>     "contents": [
>       {"role": "user", "parts": [{"text": "In three sentences, explain what a database index is and why it slows down writes."}]}
>     ],
>     "generationConfig": {
>       "thinkingConfig": {"thinkingLevel": "high"}
>     }
>   }'
> ```

> [tab] Python
>
> ```python theme={null}
> import os
> from openai import OpenAI
>
> client = OpenAI(
>     api_key=os.environ["LAOZHANG_API_KEY"],
>     base_url="https://api.laozhang.ai/v1",
>     timeout=600.0,
>     max_retries=0,
> )
> response = client.chat.completions.create(
>     model="gemini-4-carbon",
>     reasoning_effort="high",
>     messages=[{"role": "user", "content": "In three sentences, explain what a database index is and why it slows down writes."}],
> )
> print(response.choices[0].message.content)
> print(response.usage)
> ```

* OpenAI-compatible responses put the answer in `choices[0].message.content` and the thinking token count in `usage.completion_tokens_details.reasoning_tokens`.
* Gemini-native responses put the answer in `candidates[0].content.parts[].text` and the thinking token count in `usageMetadata.thoughtsTokenCount`; `modelVersion` reads `gemini-4-carbon`.
* Code that already calls Gemini 4 Argon only needs `gemini-4-argon` replaced with `gemini-4-carbon` in `model` or the URL. The API key and other parameters stay the same.
* For images, streaming and multi-turn requests in the native format, see [Gemini protocol](https://docs.laozhang.ai/en/api-reference/gemini.md). For tools, see [Function calling](https://docs.laozhang.ai/en/api-capabilities/function-calling.md).

## Thinking levels

| Endpoint | Parameter | Values |
| - | - | - |
| Gemini native | `generationConfig.thinkingConfig.thinkingLevel` | `low`, `medium`, `high` |
| OpenAI-compatible | `reasoning_effort` | `low`, `medium`, `high` |

* Leave it out and Google's default, `medium`, applies.
* The Gemini-native endpoint rejects `minimal`; the lowest level is `low`.
* Higher levels produce more thinking tokens, so they cost more and take longer. Use `low` for simple questions and `high` for hard ones.
* Google states that Gemini 4 Carbon ignores `temperature`, `top_p` and `top_k`. Sending them does not cause an error, but they have no effect.

## Who this affects

* **Applies to**: projects that call Gemini through LaoZhang API for complex coding, debugging, math and science reasoning, or multi-step agent work. Any `default`-group API key with the Usage first or Usage-based billing mode can call it.
* **Not affected**: no existing model changed with this launch. `gemini-4-argon`, `gemini-3.8-flash` and the other Gemini models keep working, and existing integrations need no changes.

## FAQ

### What is the Gemini 4 Carbon model ID?

`gemini-4-carbon`. Google published only this one stable ID, with no `-preview` or dated variant. Put it in the `model` field for the OpenAI-compatible endpoint, or in the URL path for the Gemini-native endpoint.

### Do I need a Google account or a paid Gemini API plan?

No. Google offers Gemini 4 Carbon only on the paid tier of its own API; the free tier cannot call it. On LaoZhang API you use your LaoZhang API key with the base URL `https://api.laozhang.ai`, prepaid and billed per token.

### Will the price change after December 1?

Yes. Google's introductory price ends on November 30. From December 1, Google charges \$4 input, \$20 output and \$0.2 cache read per 1M tokens. LaoZhang API will change on the same date and post a notice in [Announcements](https://docs.laozhang.ai/en/changelog.md). Actual charges are always in the [usage logs](https://api.laozhang.ai/log).

### Is Gemini 4 Carbon or Gemini 4 Argon better value?

The per-token price is the same, but Carbon spends more thinking tokens per turn, so the same request usually costs more and runs longer than on Argon. Pay for Carbon on hard problems where the better answer matters; use Argon for routine work. The comparison table above, "Gemini 4 Carbon vs Gemini 4 Argon", lists the differences.

### Does it accept images, PDFs, audio or video?

Image input works: use `image_url` (a URL or Base64) on the OpenAI-compatible endpoint, or `inlineData` / `fileData` on the Gemini-native endpoint. Google also lists video, audio and PDF as input types; see [Image understanding](https://docs.laozhang.ai/en/api-capabilities/vision-understanding.md) and [Gemini protocol](https://docs.laozhang.ai/en/api-reference/gemini.md) for the request format. The model only returns text; it does not generate images or audio.

### What if long requests time out?

With the `high` thinking level, a single request can take more than a minute. Set your client timeout to around 600 seconds and switch to streaming:

* OpenAI-compatible: add `"stream": true` to the request.
* Gemini native: call `streamGenerateContent?alt=sse` instead.

## Sources and related pages

* [Google: Gemini 4 Carbon model page](https://ai.google.dev/gemini-api/docs/models/gemini-4-carbon): model ID, context window, output limit, input types, thinking levels and supported tools.
* [Google: Gemini API pricing](https://ai.google.dev/gemini-api/docs/pricing): introductory price, the price from December 1, and implicit caching.
* [Google: Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog): general availability on November 10, 2026.
* [Models & Pricing](https://docs.laozhang.ai/en/models.md#vendor-google-text): current LaoZhang API prices.
* [Gemini 4 Argon launch announcement](https://docs.laozhang.ai/en/announcements/gemini-4-argon-2026-10.md)
* [Text generation API: protocols and common models](https://docs.laozhang.ai/en/api-capabilities/text-generation.md)
* [Gemini protocol: generateContent](https://docs.laozhang.ai/en/api-reference/gemini.md)
* [LaoZhang API announcements](https://docs.laozhang.ai/en/changelog.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.