Benchmark · general

MRCR

2 results 2 models

OpenAI-style long-context Multi-Round Co-reference Resolution; retrieval accuracy over long conversations in %.

0 25 50 75 100 2025-04-15 2025-12-24 2026-09-04 Gemini 2.0 Flash · 70.5 · 2025-04-15 Muse Spark 1.3 · 98.5 · 2026-09-04
Gemini 2.0 Flash Muse Spark 1.3
Timeline
Date Model Score Source
2026-09-04 Muse Spark 1.3 98.5% Meta releases Muse Spark 1.3 with fewer tool calls and tokens
2025-04-15 Gemini 2.0 Flash 70.5% Google releases Gemini 2.0 Flash with enhanced quality and twice the speed of Gemini 1.5 Pro