---
title: "Claude Desktop Gateway mode & API key setup"
description: "Run Claude Desktop on a CodingPass plan from $10/mo using Third-Party Inference (Gateway mode). Sonnet 5.5, Opus 5.5 and Fable 5.2 with MCP, Artifacts and prompt caching."
url: https://codingpass.ai/tools/claude-desktop
language: en
updated: 2026-11-18
---
# Claude Desktop via Gateway mode.

Enable Developer Mode, switch Third-Party Inference to Gateway mode, and run Claude Sonnet 5.5, Opus 5.5 or Fable 5.2 inside Claude Desktop on a fixed monthly plan—no claude.ai login or per-token surprise bills.

- Setup in 60s
- No surprise charges
- Full refund if you've used 5% or less

Developer → Configure Third-Party Inference

```
Inference provider   Gateway
Credential kind      Static API key
Gateway base URL     https://codingpass.ai
Gateway API key      cp-live-…
Gateway auth scheme  bearer
```

Model ID for your plan

**Go**

claude-sonnet-5-5

**Plus**

claude-opus-5-5

**Pro**

claude-fable-5-2

## Anthropic’s desktop app, on your own gateway.

Chat, Artifacts, Projects and local MCP servers—all routed through Claude Desktop’s built-in Gateway mode.

### Built-in Gateway mode

Under Developer Mode, Claude Desktop lets you set Inference provider to Gateway. No proxy scripts, no client patches and no separate Claude Pro or Max login required.

### Chat, Artifacts and Code in one window

Use Claude Desktop’s native UI—live Artifacts, file uploads, Projects and the Code tab—backed by Sonnet 5.5, Opus 5.5 or Fable 5.2 on your CodingPass plan.

### Local MCP servers work as usual

Your local MCP servers in `claude_desktop_config.json`—filesystems, Git, databases and custom tools—keep working when inference runs through CodingPass.

### Extended thinking and prompt caching

Because Claude Desktop talks to CodingPass over the native Anthropic Messages API, extended thinking, vision and prompt caching work out of the box.

## Which plan for Claude Desktop?

Every plan works with Claude Desktop’s Gateway mode. Pick the Claude flagship you want; higher tiers include everything below them.

- [Full plan details](https://codingpass.ai/pricing#compare-plans)
- [Allowances for all 27 models](https://codingpass.ai/models)

### Go: Claude Sonnet 5.5

Best for Daily chat, coding, writing and MCP workflows

$10/mo

$60 allowance a month

**Requests / month**

≈2,050

**Per 5-hour window**

≈342req

**Model ID**

claude-sonnet-5-5

### Plus: Claude Opus 5.5

Best for Large codebases, complex Artifacts, Opus 5.5

$30/mo

$200 allowance a month

**Requests / month**

≈3,420

**Per 5-hour window**

≈570req

**Model ID**

claude-opus-5-5

### Pro: Claude Fable 5.2

Best for Heavy daily use and the hardest reasoning tasks

$100/mo

$500 allowance a month

**Requests / month**

≈5,090

**Per 5-hour window**

≈848req

**Model ID**

claude-fable-5-2

Estimates from CodingPass traffic: about 100K tokens per request at roughly 5% input, 1% output and 94% cache read. Long desktop sessions with MCP tools send multiple requests. Allowances are metered at official API prices.

## Three steps. About a minute.

Turn on Developer Mode, fill in the Gateway form once, and relaunch Claude Desktop.

1. ### Create a key
   Sign in with Google, GitHub or email and pick a plan. Then go to Dashboard → Keys → New key. Keys start with `cp-live-`.
2. ### Enable Developer Mode
   You don’t need to sign in to claude.ai. Confirm the dialog, and a Developer menu appears in the menu bar.
   `Help → Troubleshooting → Enable Developer Mode…`![Claude Desktop Help menu with the Troubleshooting submenu open, listing Enable Developer Mode](https://codingpass.ai/assets/img/tools/claude-desktop/dev-mode-menu.webp)
3. ### Fill in the Gateway form
   Open Developer → Configure Third-Party Inference…. Gateway is already selected; fill in the four fields below. The base URL has no `/v1`.
   `Credential kind Static API keyGateway base URL https://codingpass.aiGateway API key cp-live-…Gateway auth scheme bearer`![Configure third-party inference window: Gateway selected, Credential kind Static API key, base URL https://codingpass.ai, a masked API key and auth scheme bearer](https://codingpass.ai/assets/img/tools/claude-desktop/gateway-form.webp)
4. ### Apply and relaunch
   Claude Desktop reopens with `Gateway` in the bottom-left corner, and the Claude models on your plan are in the model picker. Or import in one click from the CodingPass Keys page using [CC Switch](https://codingpass.ai/tools/cc-switch).
   `Apply Changes → Relaunch now`![Claude Desktop in Gateway mode with a reply in the chat and the model picker open, listing Sonnet 5.5, Opus 5.5, Fable 5.2, Haiku 5.5 and Fable 5.1](https://codingpass.ai/assets/img/tools/claude-desktop/model-picker.webp)

[Step-by-step guide with screenshots](https://codingpass.ai/blog/how-to-use-third-party-api-in-claude-desktop-gateway)

Tested with the latest stable release of Claude Desktop 6.6395 on Nov 18, 2026.

![Claude Desktop Help menu with the Troubleshooting submenu open, listing Enable Developer Mode](https://codingpass.ai/assets/img/tools/claude-desktop/dev-mode-menu.webp)

## Get more out of every allowance.

One monthly allowance covers every model, with 5-hour and weekly caps. The right model for the task makes it go further.

### Pick the right Claude model in the model picker

In Gateway mode, Claude Desktop reads `/v1/models` and lists every `claude-*` model on your plan. Use **Haiku 5.5** for quick edits and summaries, **Sonnet 5.5** for everyday work, and **Opus 5.5** (Plus) or **Fable 5.2** (Pro) when you need deeper reasoning.

```
claude-haiku-5-5    # fast, lowest cost per message
claude-sonnet-5-5   # everyday coding and writing
claude-opus-5-5     # deep reasoning, on Plus
```

### One-click setup and switching with CC Switch

If you also use Claude Code or want to switch between CodingPass and a personal Claude login, click **Import to CC Switch** on the CodingPass Keys page. CC Switch configures Claude Desktop’s Gateway settings for you and lets you toggle providers in one click.

```
Dashboard → Keys → Claude Desktop → Import to CC Switch
```

## Questions about Claude Desktop.

**Is this allowed? Could my account get banned?**

You use only your CodingPass key. You never log in to, share or relay anyone's account, so your own Claude or ChatGPT account isn't involved. Your requests follow each vendor's usage policy.

**What is Gateway mode in Claude Desktop, and why use it?**

Gateway mode (`Developer → Configure Third-Party Inference → Gateway`) is Claude Desktop’s built-in option to route model requests through your own Anthropic-compatible endpoint instead of a personal `claude.ai` subscription or Bedrock/Vertex. Pointing it at `https://codingpass.ai` with a `Bearer` key keeps your usage on a fixed monthly CodingPass plan with no surprise overage bills.

**Why don’t the models show up after saving?**

Save Changes only stores the form; click **Apply Changes**, then **Relaunch now** (or quit with `Cmd+Q` and reopen). Then check that **Credential kind** is `Static API key`, **Gateway base URL** is `https://codingpass.ai` without `/v1`, and **Gateway auth scheme** is `bearer`.

**Should Gateway base URL include /v1?**

No. Enter `https://codingpass.ai`. Claude Desktop appends `/v1/models` and `/v1/messages` itself—adding `/v1` sends requests to `/v1/v1/messages` and fails.

**Why do I only see Claude models and not GPT, DeepSeek or Kimi?**

Claude Desktop queries `/v1/models` from the Gateway and filters the picker to `claude-*` model IDs. Only the Claude models included in your CodingPass plan appear in Claude Desktop. To run GPT-6.1, DeepSeek or GLM on the same plan and key, connect [OpenCode](https://codingpass.ai/tools/opencode), [ZCode](https://codingpass.ai/tools/zcode) or [Codex](https://codingpass.ai/tools/codex).

**Do MCP servers, Artifacts, Extended Thinking and Prompt Caching still work?**

Yes. Local MCP servers in `claude_desktop_config.json`, Artifacts, file attachments, Extended Thinking and Prompt Caching all work over the Gateway. Cache reads are metered at the official cache-read price, keeping multi-turn sessions economical.

**Can I switch back to my Claude Pro or Max login?**

Yes, but Claude Desktop has no one-click toggle for it yet ([anthropics/claude-code#56606](https://github.com/anthropics/claude-code/issues/56606)). The simplest way is [CC Switch](https://codingpass.ai/tools/cc-switch): pick the official Claude Desktop entry and relaunch. Your claude.ai chats are stored separately and are still there.

**What happens when I hit the 5-hour cap?**

Each 5-hour window allows 1/6 of the monthly allowance (on Go, $10 of Sonnet 5.5, about 342 requests). The cap counts all models together, so CodingPass then returns 429 with the reset time for every model. Wait for the reset, buy a reset card for the current window, or upgrade. [How caps and 429s work](https://codingpass.ai/usage-limits#response).

## Same key, every agent.
