Inception gives new accounts 100M free tokens to try Mercury 3, no card needed

Curious how fast a diffusion LLM really is? Inception just released Mercury 3, which outputs more than 1,000 tokens per second and has a 1M-token context window. New accounts come with 100 million free tokens and no card is required, so you can start calling it as soon as you create an API key.

What you get

• Every new Inception Platform account gets 100 million free tokens, with no payment details needed at sign-up.
• The free tier allows 1,000 requests, 1M input tokens and 100K output tokens per minute, which is plenty for personal testing and small projects.
• When your free tokens run low, you can add a payment method on the Billing page and keep going on pay-as-you-go.
• Mercury 3 is 50% off at launch: $0.10 per 1M input tokens and $0.35 per 1M output tokens (regularly $0.20 and $0.70). Cached input is another 90% cheaper.

Mercury 3 specs

• Mercury 3 is a diffusion-based LLM that runs at about 1,068 tokens per second on NVIDIA GPUs.
• It has a 1M-token context window and up to 131,072 output tokens per request.
• It supports tool calling and structured outputs, with instant, low, medium and high reasoning effort.
• The API is OpenAI-compatible, so you only need to change the base URL and model name in your existing code.
• In Inception's published benchmarks, Mercury 3 matches or beats speed-focused models such as GPT-6 Luna and Claude Haiku 5.5 while running 4 to 8 times faster.

How to use it

1. Sign up on the Inception Platform. The 100 million free tokens are added to your account automatically.
2. Create an API key on the API Keys page.
3. Set the base URL to https://api.inceptionlabs.ai/v1 and the model to mercury-3, and you're ready to go.

See the Inception quick start and the Mercury 3 announcement for details.