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

# gpt-5-5

> OpenAI's GPT-5.5 — current flagship of the GPT-5 family. Vision-capable, reasoning-tunable via `reasoning_effort`. The right default for non-Anthropic main brain work or for DOSIA `chat_with` external-LLM calls.

`gpt-5-5` is the current flagship of OpenAI's GPT-5 family. Vision-capable on image input, reasoning depth tunable via `reasoning_effort`, broad tool / function support, JSON mode and structured outputs. On the gateway, it's the default landing spot when DOSIA's `text-writing-tools.chat_with` is asked to "use a different LLM" — so a user prompt like "have GPT rewrite this paragraph" routes here.

**Pricing:** $5.00 / 1M input, $30.00 / 1M output, \$0.50 / 1M cache read — see the [rate card](/pricing). Failures don't bill.

## Protocols

| Protocol                | Path                                                |
| ----------------------- | --------------------------------------------------- |
| OpenAI Chat Completions | `POST https://llm.bytespike.ai/v1/chat/completions` |

## Quickstart

<CodeGroup>
  ```bash cURL theme={null}
  curl https://llm.bytespike.ai/v1/chat/completions \
    -H "Authorization: Bearer $BYTESPIKE_API_KEY" \
    -H "content-type: application/json" \
    -d '{
      "model": "gpt-5-5",
      "messages": [
        { "role": "user", "content": "Untangle this contract paragraph." }
      ],
      "reasoning_effort": "medium"
    }'
  ```

  ```python Python (openai SDK) theme={null}
  from openai import OpenAI

  client = OpenAI(
      base_url="https://llm.bytespike.ai/v1",
      api_key=os.environ["BYTESPIKE_API_KEY"],
  )

  resp = client.chat.completions.create(
      model="gpt-5-5",
      messages=[{"role": "user", "content": "Untangle this contract paragraph."}],
      reasoning_effort="medium",
  )

  print(resp.choices[0].message.content)
  ```

  ```typescript TypeScript (openai SDK) theme={null}
  import OpenAI from "openai"

  const client = new OpenAI({
    baseURL: "https://llm.bytespike.ai/v1",
    apiKey: process.env.BYTESPIKE_API_KEY,
  })

  const resp = await client.chat.completions.create({
    model: "gpt-5-5",
    messages: [{ role: "user", content: "Untangle this contract paragraph." }],
    reasoning_effort: "medium",
  })

  console.log(resp.choices[0].message.content)
  ```
</CodeGroup>

## Capabilities

| Capability                             | Supported                  |
| -------------------------------------- | -------------------------- |
| Chat Completions                       | ✅                          |
| Streaming (SSE)                        | ✅                          |
| Vision (image input)                   | ✅                          |
| Tools / function calling               | ✅ parallel                 |
| JSON mode                              | ✅                          |
| Structured outputs                     | ✅                          |
| `reasoning_effort` low / medium / high | ✅                          |
| Context window                         | 128K tokens                |
| Modality                               | chat + vision              |
| Capability bucket                      | `vision` + `external_chat` |

## When to use

* **Default GPT-5 family pick** — when "use GPT-5" is the brief, this is the alias.
* **DOSIA `chat_with`** — main brain calls this when the user asks for a non-Claude voice ("have GPT rewrite this").
* **Reasoning-heavy chat** — pair with `reasoning_effort: "high"` for harder analysis.

When **not** to use:

* High-volume / budget-bound — drop to [`gpt-5-4-mini`](/models/gpt-5-4-mini).
* You want Claude's `tool_use` / `cache_control` block model — go to [`claude-sonnet-4-6`](/models/claude-sonnet-4-6).

## Next

* [`gpt-5-4`](/models/gpt-5-4) — previous GPT-5 family flagship
* [`claude-opus-4-8`](/models/claude-opus-4-8) — Anthropic alternative for hard reasoning
* [DOSIA MCP integration](/concepts/mcp-integration) — `chat_with` tool surface