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

# Fire Pass

> Set up Fire Pass to access open-weight model routers like kimi-fast-latest for agentic coding in Claude Code, OpenCode, Cline, Kilo Code, and LangChain.

Fire Pass gives eligible users access to selected open-weight model routers for personal, non-production agentic coding. Included requests are covered while the pass is active, subject to its limits and terms. Check the [Fire Pass page](https://app.fireworks.ai/fire-pass) for your status and currently enabled models.

> [!NOTE]
>
> Fire Pass is experimental and is not currently open to new subscriptions.
> Existing users and promo recipients can continue to manage their pass.
> Features, availability, and pricing are subject to change.

## What's included

With an active Fire Pass you get:

* Covered usage on the open-weight model routers enabled for your pass
* A dedicated `fpk_...` key that works only with Fire Pass routes
* Access through compatible OpenAI- or Anthropic-style clients

Use a standard Fireworks key (`fw_...`) for models and features outside Fire Pass.

## Getting Started

**1. Create an account**

If you haven't already, sign up for a [Fireworks
account](https://app.fireworks.ai).

**2. Activate your pass**

Go to the [**Fire Pass** page](https://app.fireworks.ai/fire-pass). If you
received a promo code, enter it there and review the pass details.

**3. Get a Fire Pass API key**

Create a dedicated Fire Pass API key. Requests to enabled routes are covered
while your pass is active, subject to its limits and terms.

## Using your Fire Pass

Fire Pass is designed for personal agentic coding. Configure your tool manually with your Fire Pass key and the details in [General Configuration](#general-configuration).

> [!WARNING]
>
> [Nexus CLI](https://fireworks.ai/docs/nexus/cli.md) doesn't accept Fire Pass keys. It rejects
> an `fpk_...` key and asks you to sign in with `nexus login`. To use
> Nexus CLI, sign in with a standard Fireworks key (`fw_...`).

> [!WARNING]
>
> Fire Pass keys cannot use `auto` or Nexus Auto. Codex and the
> ChatGPT desktop app use the Responses API, which isn't supported with Fire
> Pass keys.

### General Configuration

If your tool supports custom API endpoints, you can configure it manually using these details:

* **Model ID**: `accounts/fireworks/routers/kimi-fast-latest`
* **API Key**: Generate a new Fire Pass API key (generated at [app.fireworks.ai/api-keys](https://app.fireworks.ai/api-keys))
* **Base URL (OpenAI-compatible)**: `https://api.fireworks.ai/inference/v1`
* **Base URL (Anthropic-compatible)**: `https://api.fireworks.ai/inference` (some tools may expect `/v1/messages` appended)

`kimi-fast-latest` currently resolves to Kimi K3 Fast, with a 1,040,000-token
context window and up to 131,072 output tokens. Recheck the [Fire Pass
page](https://app.fireworks.ai/fire-pass) when the latest alias changes.

> [accordion] OpenClaw
>
> Add Fireworks as an OpenAI-compatible provider in
> `~/.openclaw/openclaw.json`. Use the base URL, Fire Pass key, and
> `kimi-fast-latest` model ID from [General Configuration](#general-configuration).
>
> > [!WARNING]
> >
> > Do not share your Fire Pass key or commit it to version control.
>
> Follow the [OpenClaw provider documentation](https://openclaw.ai) to finish
> configuring and testing the custom provider.

> [accordion] OpenCode
>
> Add the Fire Pass router to `~/.config/opencode/opencode.json`, with your
> Fire Pass key in `FIREWORKS_API_KEY`:
>
> ```json ~/.config/opencode/opencode.json theme={null}
> {
>   "$schema": "https://opencode.ai/config.json",
>   "provider": {
>     "fireworks-ai": {
>       "options": { "apiKey": "{env:FIREWORKS_API_KEY}" },
>       "models": {
>         "accounts/fireworks/routers/kimi-fast-latest": { "name": "Kimi Fast (Fire Pass)" }
>       }
>     }
>   },
>   "model": "fireworks-ai/accounts/fireworks/routers/kimi-fast-latest"
> }
> ```

> [accordion] Cline
>
> To configure Cline with your Fire Pass:
>
> 1. Install the Cline extension in VS Code from [cline.bot](https://cline.bot) or the VS Code marketplace
> 2. Open VS Code settings and search for "Cline" or use the Cline panel
> 3. Configure the following settings:
>
> **API Configuration:**
>
> * **API Provider**: OpenAI Compatible
> * **Base URL**: `https://api.fireworks.ai/inference/v1`
> * **OpenAI Compatible API Key**: Your Fire Pass key (from [app.fireworks.ai/api-keys](https://app.fireworks.ai/api-keys))
> * **Model ID**: `accounts/fireworks/routers/kimi-fast-latest`
>
> **Model Configuration:**
>
> * **Supports Images**: Enabled (checked)
>
> * **Context Window Size**: 1,040,000
>
> * **Max Output Tokens**: 131,072
>
> * **Input Price / 1M tokens**: 0 while covered by an active Fire Pass
>
> * **Output Price / 1M tokens**: 0 while covered by an active Fire Pass
>
> 4. Save your settings and send a request. Fire Pass applies when the key and route are eligible.

> [accordion] Claude Code
>
> Add these values to `~/.claude/settings.json`:
>
> ```json ~/.claude/settings.json theme={null}
> {
>   "env": {
>     "ANTHROPIC_BASE_URL": "https://api.fireworks.ai/inference",
>     "ANTHROPIC_AUTH_TOKEN": "fpk_YOUR_FIRE_PASS_KEY",
>     "ANTHROPIC_DEFAULT_OPUS_MODEL": "accounts/fireworks/routers/kimi-fast-latest",
>     "ANTHROPIC_DEFAULT_SONNET_MODEL": "accounts/fireworks/routers/kimi-fast-latest",
>     "ANTHROPIC_DEFAULT_HAIKU_MODEL": "accounts/fireworks/routers/kimi-fast-latest",
>     "ANTHROPIC_DEFAULT_FABLE_MODEL": "accounts/fireworks/routers/kimi-fast-latest",
>     "CLAUDE_CODE_SUBAGENT_MODEL": "accounts/fireworks/routers/kimi-fast-latest"
>   },
>   "model": "accounts/fireworks/routers/kimi-fast-latest"
> }
> ```
>
> Run `claude`. Claude Code may ask once whether to use the API key; approve it.
>
> Set every model slot to the Fire Pass router. Claude Code sends background
> and subagent requests to those slots, and a `claude-*` model ID fails with a
> Fire Pass key. Claude Code's web search also needs a standard `fw_...` key.

> [accordion] Kilo Code
>
> To configure Kilo Code with your Fire Pass:
>
> 1. Install the Kilo Code extension in VS Code from the [VS Code marketplace](https://marketplace.visualstudio.com/items?itemName=kilocode.Kilo-Code) or [kilocode.ai](https://kilocode.ai)
> 2. Open Kilo Code and select **Bring my own Key** on the "How would you like to get started?" screen
> 3. Configure the following settings:
>
> **API Configuration:**
>
> * **API Provider**: OpenAI Compatible
> * **Base URL**: `https://api.fireworks.ai/inference/v1`
> * **API Key**: Your Fire Pass key (from [app.fireworks.ai/api-keys](https://app.fireworks.ai/api-keys))
> * **Model**: `accounts/fireworks/routers/kimi-fast-latest`
>   * Type or paste the router ID and select "Use custom" when it appears
>
> **Model Configuration:**
>
> * **Context Window Size**: 1,040,000
>
> * **Supports images**: Enabled (checked)
>
> * **Input Price**: 0 while covered by an active Fire Pass
>
> * **Output Price**: 0 while covered by an active Fire Pass
>
> 4. Save your settings and send a request. Fire Pass applies when the key and route are eligible.

> [accordion] LangChain Deep Agent
>
> Set `FIREWORKS_API_KEY` to your Fire Pass key from [app.fireworks.ai/api-keys](https://app.fireworks.ai/api-keys). Install [LangChain Deep Agents](https://docs.langchain.com/oss/python/deepagents/models) with the Fireworks integration, then pass the latest Fire Pass router as `fireworks:accounts/fireworks/routers/kimi-fast-latest`:
>
> ```bash theme={null}
> pip install deepagents langchain-fireworks
> ```
>
> ```python theme={null}
> from deepagents import create_deep_agent
>
> agent = create_deep_agent(
>     model="fireworks:accounts/fireworks/routers/kimi-fast-latest",
>     system_prompt="You are a helpful coding assistant.",
> )
> result = agent.invoke(
>     {"messages": [{"role": "user", "content": "Hello"}]},
> )
> ```
>
> See the LangChain docs for [models](https://docs.langchain.com/oss/python/deepagents/models) and [customization](https://docs.langchain.com/oss/python/deepagents/customization).

## Terms of Use

Fire Pass is intended for **non-production coding use only**. By activating Fire Pass you agree to the following:

* **Allowed**: Personal development, experimentation, and coding with supported agentic harnesses and compatible clients
* **Prohibited**: Production workloads and any use that violates the [Fireworks Terms of Service](https://fireworks.ai/terms-of-service)

> [!WARNING]
>
> Violations of these terms may result in pass revocation.

## FAQ

> [accordion] Can I use Fire Pass with any model?
>
> No. Fire Pass covers only its enabled routes. Check the Fire Pass page for
> the latest set.

> [accordion] Do I need an invite code?
>
> Fire Pass is not currently open to new subscriptions. Existing users and
> promo recipients can manage access on the Fire Pass page.

> [accordion] Do I need to use a special model ID?
>
> Use one of the routers enabled for Fire Pass, such as
> `accounts/fireworks/routers/kimi-fast-latest`. Check the Fire Pass page for
> the current set.

> [accordion] Can I use Nexus Auto or Auto?
>
> No. `auto` requires a standard Fireworks
> key (`fw_...`). Fire Pass keys can use only Fire Pass-enabled routes.

> [accordion] How will this appear on my billing dashboard?
>
> Included usage remains visible while it is covered by your active pass.
> View the pass status and expiration date on the [**Billing**
> page](https://app.fireworks.ai/account/billing).

> [accordion] Can I use other models with my agentic harness?
>
> Yes, with a standard Fireworks key (`fw_...`). A Fire Pass key (`fpk_...`)
> works only with routes enabled for Fire Pass. Other models use standard
> per-token billing.


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