What is Inception
Inception Labs provides diffusion-based language models through a hosted API, including Mercury for general generation and Mercury Edit for code editing.
Free, Developer and Enterprise access
As of September 19, 2026, Free includes 10 million tokens and access to all current models. Developer is usage based with higher limits and priority support. Enterprise uses custom rates and adds volume pricing, security review, service-level commitments and dedicated support.
Model rates and features
Mercury 2.5 supports a 260,000-token context window and currently has a promotional rate of US$0.04 per million input tokens, US$0.004 cached-input tokens and US$0.15 output tokens; its listed rates before the promotion are US$0.20, US$0.02 and US$0.75. Mercury 2 and Mercury Edit 2 remain US$0.25 input, US$0.025 cached input and US$0.75 output per million tokens. Mercury Edit 2 provides fill-in-the-middle and NextEdit workflows for code changes.
The hosted API supports compatible chat and generation workflows, streaming and model-specific controls. Usage beyond the free allowance is charged at the selected model's published token rate.
Official sources: Inception model pricing and access tiers, Inception model catalogue.