Inception Mercury (diffusion language model): 100 million free tokens for new accounts, no payment details

Each new account gets a one-time 100 million (100M) free tokens shared across all models; the vendor states no payment details are required, and you add payment details under Billing to keep going once they run out

Inception Platform docs, Get Started: "Every new account includes 100 million free tokens — no payment details required." "Each new user is initially assigned 100 million free tokens to help get started with the API." FAQ: "Every new account includes a one-time credit of 100 million free tokens. The credit is shared across all models rather than granted per model." Free-tier rate limits (docs, Rate Limits): 1,000 requests, 1M input tokens and 100K output tokens per minute. Mercury 3 has a 1M context and up to 131,072 output tokens, supports tool calling and structured outputs, and uses an OpenAI-compatible API at https://api.inceptionlabs.ai/v1; it works with Claude Code (through LiteLLM), Cursor, Zed, Kilo Code, OpenCode and more. Current official pricing for Mercury 3 is 0.10 USD input / 0.35 USD output per million tokens (a launch-period half-price discount).

⚠️The credit is one-time and does not renew monthly; the docs give neither an expiry period nor a deadline. Mercury is not a mainstream model such as Claude, GPT or DeepSeek but Inception's own diffusion language model, known for fast generation; it takes text input only and offers no images or embeddings. The credit is given only on Inception's own platform (platform.inceptionlabs.ai); Mercury on third-party channels such as OpenRouter is the paid version. The sign-up method (email or Google) and any regional restriction were not checked through the sign-up flow; reaching overseas endpoints from mainland China may need a stable overseas network, which was not tested.