Mercury 2 is Inception Labs' diffusion-based LLM that generates responses in parallel rather than token-by-token, delivering very high throughput (1000+ tokens/second) with a 128K context window. It offers tunable reasoning and native tool use, matching Claude Haiku/Gemini Flash-class quality at much faster speeds and lower cost, ideal for latency-sensitive agentic and coding tasks.
Cost rate
0.2x
Context
128K
Released
Mar 4, 2026
Input
Text
Output
Text
Support
ReasoningTool callingStructured outputs
Performance
Median latency and throughput measured across recent requests.
Throughput512 tok/s
Latency6.49s
Latest news about Mercury 2
Recent articles and community posts, linked to their original sources.
I Tested NEW Mercury 3: Fast Diffusion LLM - But Can it Code?AI Coding Daily12,686