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

# OpenAI GPT via RuAPI: the three models and when to use each

> The versatile GPT line: gpt-5.5, gpt-5.4 and gpt-5.4-mini. What each is best for, why GPT wins on ecosystem support, and when Claude, Gemini or DeepSeek is the honest pick.

OpenAI's GPT is the model to reach for when you don't yet know exactly what you need. Broad capability across tasks, the largest tooling ecosystem, and predictable function calling make it a safe first choice for most projects. Three versions are available through RuAPI, differing mainly in the balance of "smarter versus faster and cheaper." Below is how to choose between them — and when another line is the more honest pick.

## The three GPT models

| Model            | Best for                                                                   | Speed / cost    |
| ---------------- | -------------------------------------------------------------------------- | --------------- |
| **gpt-5.5**      | Most capable: hard code, deep analysis, multi-step agents, tough reasoning | Slower, pricier |
| **gpt-5.4**      | Balanced everyday workhorse: code, chat, integrations                      | Medium          |
| **gpt-5.4-mini** | Simple tasks at high volume: classification, tagging, short replies        | Faster, cheaper |

<Tip>
  Not sure? Take **gpt-5.4** — it covers almost everything. Step up to **gpt-5.5** when you hit a quality ceiling on hard tasks, and down to **gpt-5.4-mini** when speed and cost matter more than peak intelligence.
</Tip>

Exact model IDs and per-token prices are on the [**Pricing** page at www.ruapi.ai](https://www.ruapi.ai). New versions appear in the catalog automatically.

## Versatile and well-supported

The main reason to reach for GPT isn't a peak score in one narrow area — it's that it handles almost any task with equal confidence: code, analysis, chat, extraction, summarization. You never have to remember "what this model is good at."

* **Ecosystem** — GPT is the most common backend, so nearly every library, framework and no-code tool is written against it. Whatever you integrate, the GPT path is usually the best-trodden one.
* **OpenAI-compatible protocol** — RuAPI serves the models in that same format, so any OpenAI SDK works. Only `base_url` changes.
* **Reliable function calling** — GPT consistently returns structured tool calls, the basis for agents (see [function calling](/en/function-calling) for an example).
* **Vision** — you can pass images as input: screenshots, diagrams, scans (see [image input](/en/vision)).

For long answers you can [stream the response](/en/streaming) token by token instead of waiting for the full reply.

## When to pick GPT (and when another line)

GPT is a strong default. Here's the honest picture of where it stops being the best pick:

* **Pick GPT** when you want an all-rounder, maximum tool compatibility, or you simply don't want to guess — **gpt-5.4** covers most scenarios.
* **[Claude](/en/claude-api)** is often more careful in long agentic chains and across large codebases.
* **[Gemini](/en/gemini-api)** wins on very long context and multimodal tasks with heavy input.
* **[DeepSeek](/en/deepseek-api)** is attractive when cost at scale drives the decision and hard reasoning isn't critical.

<Note>
  One key covers all lines. You can switch freely between GPT, Claude, Gemini and DeepSeek within a single project — only the `model` value changes.
</Note>

## Connecting GPT

GPT speaks the **OpenAI-compatible** protocol, so any OpenAI SDK or a tool like Cursor, Cline or n8n works. You only need to point `base_url` at RuAPI:

```python theme={null}
from openai import OpenAI

client = OpenAI(
    api_key="YOUR_RUAPI_KEY",
    base_url="https://www.ruapi.ai/v1",
)

resp = client.chat.completions.create(
    model="gpt-5.4",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(resp.choices[0].message.content)
```

Step by step, with the key and curl, in the [Quickstart](/en/quickstart). You top up in USDT and no VPN is needed; a Visa / Mastercard also works — see [Top up](/en/topup).

## Next

<CardGroup cols={2}>
  <Card title="Connect and first request" icon="bolt" href="/en/quickstart">
    base\_url, key, Python and curl code.
  </Card>

  <Card title="Claude models" icon="brain" href="/en/claude-api">
    Same key — when agent precision matters more.
  </Card>

  <Card title="Gemini models" icon="sparkles" href="/en/gemini-api">
    Very long context and multimodality.
  </Card>
</CardGroup>