Cohere: AI model benchmarks

AI models from Cohere. 22 models tracked, 22 with a unified ELO rating. Current best: command-a-reasoning-08-2025 (ELO 1609, rank #234 overall). Most recent tracked release: 2026-06-09.

Rated models

RankModelProviderUnified ELOBenchmarks
234command-a-reasoning-08-2025Cohere1609 ± 116
359North Mini CodeCohere1571 ± 140
384c4ai-command-a-03-2025Cohere1564 ± 131
423Command A+Cohere1552 ± 147
472Command ACohere1541 ± 1347
480c4ai-aya-expanse-32BCohere1540 ± 123
608c4ai-aya-expanse-8BCohere1518 ± 120
684aya-expanse-32BCohere1500 ± 1370
729c4ai-command-r-plusCohere1491 ± 142
840aya-23-35BCohere1469 ± 1333
861c4ai-command-r-08-2024Cohere1466 ± 188
903Command-R+Cohere1458 ± 179
922Command-R+ (08-2024)Cohere1454 ± 164
926c4ai-command-r-v01Cohere1453 ± 163
951c4ai-command-r7B-12-2024Cohere1448 ± 151

Interactive version: theaggregate.ai/company?slug=cohere · How It Works · Data refreshed daily, snapshot 2026-09-05.