Cohere adds world-class open models to Model Vault, its dedicated inference platform
cohere · x · 2026-11-04
Cohere announced that Model Vault now hosts as many world-class open models as it can — both its own and third-party ones — arguing open access benefits everyone. Models run on single-tenant infrastructure with no shared GPUs and no neighboring tenants, deployable on dedicated or fully encrypted infrastructure, positioning the product as SaaS-simple yet fully isolated inference for enterprise workloads.
Related event: Cohere Model Vault Adds Open-Weight Models to Encrypted Inference Platform(2 posts)→
More from Infra
- Nvidia lost credibility in Washington by exaggerating China's AI chip capabilities, CFR fellow says — mchorowitz · 2026-11-04
- llama.cpp PR Delivers Big Prompt-Processing Speedups for Gemma E2B/E4B — jacek2023 · 2026-11-04
- Baseten's Philip Kiely: agents need gateways, sandboxes and post-training, not just inference — AccBalanced · 2026-11-04
- Repeated-data pretraining favors smaller batches, finds Sewon Min's GPU-time tradeoff study — sewon__min · 2026-11-04
- Marvell to Acquire Lumentum for AI Optical Connectivity, Closing Expected H1 2028 — BenBajarin · 2026-11-04
- DeepSeek opened its AI stack: a guide to what you can actually build with it — TheTuringPost · 2026-11-04