Cohere Embed 5 splits indexing from query cost

Cohere released Embed 5 with Pro and Fast tiers at $0.12 and $0.08 per million text tokens, respectively, while both keep a 128K-token context and share one embedding space. The launch specification says teams can index with Pro and query with Fast without rebuilding the index; across 40 development datasets, the cross-model combinations averaged 1.6% and 2.7% losses against same-model baselines. High-volume RAG teams should test Pro for offline indexing and Fast on live queries, but benchmark relevance, latency, and image pricing before treating the shared space as a free cost cut; the archive’s model-routing analysis makes the same case for versioned route evidence.