OpenRouter ships Mercury 2.5 Preview — fastest reasoning LLM at ~1,107 tok/s

OpenRouter launched Mercury 2.5 Preview from Inception exclusively on its platform: a reasoning LLM with parallel token generation (~1,107 tokens/sec), tunable reasoning, parallel tool calls, and schema-aligned JSON — aimed at latency-sensitive workloads. On 8 Sep Inception formally introduced Mercury 2.5 as its production dLLM (~40% intelligence lift vs Mercury 2, 1,107 tok/s, 260K context, $0.20/$0.75 per M tokens with launch discount), plus Mercury Voice and Mercury Router previews.