Ollama publishes v0.40.5-rc1 release candidate, adding NVIDIA Nemotron 3.5 Super to its MLX engine
Ollama published a release candidate tagged v0.40.5 rc1 on Oct 15, 2026, at 10:28 AM, and its release notes announce support for NVIDIA Nemotron 3.5 Super in O…
Published on MyPrivateClaw
Oct 16, 2026, 6:42 AM UTC
Coverage date
Oct 15, 2026
Last updated
Oct 16, 2026, 6:42 AM UTC
News summary
Ollama published a release candidate tagged v0.40.5 rc1, by @dhiltgen on Oct 15, 2026, at 10:28 AM, for the Ollama model server/inference runtime project (PR 19030). The tag carries release candidate status, not a stable final release, and the final v0.40.5 release notes and versioning may differ in content. The v0.40.5 rc1 release notes announce support for a new model, adding NVIDIA Nemotron 3.5 Super to Ollama's MLX engine. The primary release notes do not state parameter counts, active parameters, or benchmark figures for NVIDIA Nemotron 3.5 Super. The v0.40.5 rc1 changelog lists mlx Nemotron work: accepting published Nemotron layer names while retaining legacy aliases, fixing layer count inference and detached MTP quantization, stabilizing Mamba softplus, and supporting per expert F32 global scales and ModelOpt/compressed tensors NVFP4 imports. The same changelog lists a ci change…