Skip to main content
Developer Pass is an invite-only weekly pass that gives you access to Kimi K2.5 Turbo for use in personal agentic coding harnesses like OpenCode, Cline, Kilo Code, and OpenClaw — with no per-token charges. Kimi K2.5 Turbo is a private preview of a faster Kimi K2.5 serverless API.
Beta Feature: Developer Pass is currently an experimental product in beta. Features, availability, and pricing are subject to change.

​
What’s Included

With an active Developer Pass you get:
  • Kimi K2.5 Turbo requests with zero per-token costs (a powerful reasoning model with a 256k context window, optimized for complex coding tasks)
  • Access through the standard Fireworks API using your existing API key
  • Full compatibility with OpenAI- and Anthropic-compatible agentic coding tools
Only Kimi K2.5 Turbo is covered by the Developer Pass. Usage of regular Kimi K2.5, or any other model on Fireworks, will continue to incur standard per-token charges.

​
Pricing

Cost$20 per week
Payment MethodBilled directly to your credit card on file (does not consume API credits)
Auto-renewOptional — you can enable auto-renewal, but renewal is not guaranteed
AvailabilityInvite-only

​
Getting Started

1

Create an account

If you haven’t already, sign up for a Fireworks account and generate an API key on the API Keys page.
2

Get an invite code

Developer Passes are currently invite-only. You’ll receive an invite code from Fireworks or through a partner program.
3

Redeem your code

Go to the Billing page and enter your invite code. This unlocks the Developer Pass details and purchase option.
4

Purchase your pass

Review the pass details and complete your purchase. Your pass activates immediately and is valid for one week.
5

Start coding

Use your existing Fireworks API key with any of the supported agentic harnesses below. Requests to Kimi K2.5 Turbo will no longer incur per-token charges while your pass is active.

​
Using Your Developer Pass

The Developer Pass is designed for use with personal agentic coding harnesses. Below are setup guides for each supported tool.

​
General Configuration

If your tool supports custom API endpoints, you can configure it manually using these details:
  • Model ID: accounts/fireworks/routers/kimi-k2p5-turbo
  • API Key: Your standard Fireworks API key (generated at 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)
OpenCode supports multiple config locations. Add this to any supported config location (such as ~/.config/opencode/opencode.json or opencode.json in your project):
{
  "$schema": "https://opencode.ai/config.json",
  "model": "fireworks-ai/accounts/fireworks/routers/kimi-k2p5-turbo",
  "provider": {
    "fireworks-ai": {
      "options": {
        "apiKey": "<your-fireworks-api-key>"
      },
      "models": {
        "accounts/fireworks/routers/kimi-k2p5-turbo": {
          "id": "accounts/fireworks/routers/kimi-k2p5-turbo",
          "name": "Kimi K2.5 Turbo (Developer Pass)",
          "family": "kimi",
          "attachment": false,
          "reasoning": true,
          "tool_call": true,
          "temperature": true,
          "modalities": {
            "input": ["text", "image"],
            "output": ["text"]
          },
          "limit": {
            "context": 256000,
            "output": 256000
          }
        }
      }
    }
  }
}
See OpenCode config documentation for all config locations and precedence rules.
To configure Cline with your Developer Pass:
  1. Install the Cline extension in VS Code from 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 Fireworks API key (from app.fireworks.ai/api-keys)
  • Model ID: accounts/fireworks/routers/kimi-k2p5-turbo
Model Configuration:
  • Supports Images: Enabled (checked)
  • Context Window Size: 256000
  • Max Output Tokens: 256000
  • Input Price / 1M tokens: 0 (Developer Pass covers all costs)
  • Output Price / 1M tokens: 0 (Developer Pass covers all costs)
  • Temperature: 1
  1. Save your settings and start using Cline — your Developer Pass will automatically apply to requests to the turbo router
To configure Roo Code with your Developer Pass:
  1. Install the Roo Code extension in VS Code from the VS Code marketplace or roocode.com
  2. Open Roo Code and click the gear icon to access settings
  3. On the “Choose your provider” screen, select 3rd-party Provider
  4. From the API Provider dropdown, select Fireworks AI
  5. Enter your Fireworks API key (from app.fireworks.ai/api-keys)
  6. In the Model field, type or paste the custom router ID: accounts/fireworks/routers/kimi-k2p5-turbo
    • Since this is a specialized router, you’ll see a “Use custom” option appear—select it
  7. Save your settings and start using Roo Code — your Developer Pass will automatically apply to requests to the turbo router
To configure Kilo Code with your Developer Pass:
  1. Install the Kilo Code extension in VS Code from the VS Code marketplace or 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 Fireworks API key (from app.fireworks.ai/api-keys)
  • Model: accounts/fireworks/routers/kimi-k2p5-turbo
    • Type or paste the router ID and select “Use custom” when it appears
Model Configuration:
  • Context Window Size: 256000 (256K tokens)
  • Supports images: Enabled (checked)
  • Input Price: 0 (Developer Pass covers all costs)
  • Output Price: 0 (Developer Pass covers all costs)
  1. Save your settings and start using Kilo Code — your Developer Pass will automatically apply to requests to the turbo router
Coming soon — setup instructions for using Fireworks with OpenClaw.

​
Terms of Use

The Developer Pass is intended for personal agentic coding use only. By purchasing a Developer Pass you agree to the following:
  • Allowed: Personal development, experimentation, and coding with agentic harnesses (OpenCode, Cline, Kilo Code, OpenClaw, and similar tools)
  • Prohibited: Production workloads, team or shared usage, and any use that violates the Fireworks Terms of Service
Violations of these terms may result in pass revocation without refund.

​
FAQ

No. The Developer Pass covers Kimi K2.5 Turbo only. All other models, including regular Kimi K2.5, are billed at standard per-token rates.
When your pass expires, Kimi K2.5 Turbo requests will be billed at standard per-token rates. If you have auto-renew enabled and a pass is available, it will renew automatically.
No. Auto-renewal is best-effort and your renewal may not go through.
No. The Developer Pass is for personal use only. Team or shared usage is prohibited under the Terms of Service.
Invite codes are distributed by Fireworks directly or through partner programs. There is currently no self-serve way to obtain an invite code.
Yes. You must use the specific router ID for the turbo model: accounts/fireworks/routers/kimi-k2p5-turbo. The Fireworks billing system will automatically detect your active Developer Pass and zero out the cost for requests to this endpoint.
Your usage of Kimi K2.5 Turbo will still be logged in your dashboard so you can track your request volume, but the associated cost will show as $0.00 while your pass is active. You can also view your pass’s active status and expiration date directly on the Billing page.
You can turn off auto-renewal at any time from the Billing page in your Fireworks dashboard. Your pass will remain active until the end of the current billing week.
Yes. You can use any Fireworks model with your agentic harness. However, only Kimi K2.5 Turbo requests are covered by the Developer Pass. Usage of other models will be billed to your account at standard per-token rates.