Cohere AI Updates: October 1, 2026

1. Embed 5 Launches in Pro and Fast Variants That Share One Embedding Space

Cohere. Cohere released Embed 5 as two models, embed-v5.0-pro for retrieval quality and embed-v5.0-fast for low-latency queries, both supporting text, images, and mixed text-image inputs across 100+ languages with a 128K-token context window and output dimensions from 256 to 2048 (float, int8, or binary). Because Pro and Fast share an embedding space, teams can index a corpus with Pro and query it with Fast without re-indexing; Cohere reports Fast at about 2.4x Pro’s throughput (377.3 vs 159.7 documents per second). Pricing is $0.12 per million text tokens for Pro and $0.08 for Fast, with images at $0.40 per million tokens, and the models are generally available on the Cohere API, Model Vault, Microsoft Foundry, and Amazon SageMaker. Source

2. Cohere Reports Leading Scores on Visual Document and Financial Retrieval

Cohere. On ViDoRe V3, Cohere reports Embed 5 Pro at 85.8 and Fast at 84.5, compared with Voyage 4 Large at 83.7, Gemini Embedding 2 at 83.2, and OpenAI text-embedding-3-large at 75.5. Pro also posted 80.1 on FinanceBench, 90.0 on FinQA, and an 84.8 average on parsed PDFs, with multilingual gains of 12 to 13 points in Farsi, Telugu, and Hindi over the prior generation. These are vendor-reported figures, so teams evaluating a migration from Embed 4 should benchmark on their own corpora. Source