Provider pricing
Inception Labs AI API pricing
Inception Labs ships Mercury Coder, the first commercial diffusion LLM. Generates code 5-10× faster than autoregressive transformers at comparable quality by denoising entire token blocks in parallel.
Founded2024
HeadquartersPalo Alto, CA
Tracked modelsDirectory
Best for
Code generation and agent loops where throughput dominates cost.
Strengths
- Diffusion-based generation — 1,000+ tokens/sec
- Parallel block decoding vs. sequential autoregression
- Mercury Coder competitive with GPT-4o-mini on HumanEval
- Drop-in OpenAI-compatible API