Skip to content

Ollama 0.40.11 adds macOS speech and d3 decisions

Updated Ollama

Summary

Ollama adds streamed speech generation on macOS and d3 models for classification and probability-based decisions.

Tool
Ollama
Category
Agent infra
Date
Pricing
Not stated in the sources we cited.

Ollama published v0.40.11 on October 24, 2026.

Speech on macOS

The release says Ollama can now generate speech on macOS. The command-line interface plays supplied text or a chat model's reply aloud while audio is generated. Applications can stream audio through the OpenAI-compatible /v1/audio/speech endpoint.

The notes demonstrate this with the phonon model. They do not establish speech support on every operating system or every model, so the macOS qualification matters.

Decision models

The release also brings vLLM Semantic Router's d3 decision models to Ollama. These accept text, structured state or an image for classification, yes/no questions or scoring options with probabilities. The listed family sizes are 2B, 4B, 9B and 27B.

These capabilities expand the kinds of applications Ollama can host; they are not evidence that a particular model is more accurate or faster than an alternative. Model-weight licenses and suitable local hardware remain separate questions from Ollama's own code license. See the Ollama hub.

More from Ollama

View all

Questions

What is Ollama?

According to its site, Ollama gives access to the latest open models with privacy, locally or in the cloud. The repository describes it as a way to get up and running with models such as Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen and Gemma.

Is Ollama open source?

The GitHub API reports the ollama/ollama repository license as MIT.

What is the latest launch tracked here?

Ollama v0.40.2, published October 8, 2026, which upgrades previously downloaded models in the background for llama.cpp.

Filed on Agent Radar. Claims in this note are limited to the sources above.