Mercury
by Inception · Model · Language & code models
Mercury 2.5 diffusion LLM runs at 1,107 tokens/s on widely available NVIDIA GPUs, with 260K context
Evidence
- Nov 13Inception Labs is reportedly discontinuing the Mercury models API
- Sep 8Priced $0.20/M input and $0.75/M output tokens; launch promotion 80% off; 100M free API tokens
- Sep 8Claims 40% intelligence gain over Mercury 2, comparable to GPT-5.6 Luna (Low), Gemini 3.5 Flash-Lite, Claude Haiku 4.5
- Sep 8Output reported good and very fast; a refusal classifier blocked questions about its architecture, three errors in one agent turn
- Sep 8Preview tested as usable general chatbot, comparable to some last-gen open-weights models, but worse at agentic coding with custom harness