Ollama releases: v0.32.6

TL;DR

Ollama released version v0.32.6, which includes enhancements to Qwen3.5's performance on Apple GPUs by automatically using the model's MTP head for speculative decoding and updates to `/v1/chat/completions` streaming to match OpenAI's wire format, improving compatibility.

Detailed Summary

Ollama released version v0.32.6, which includes improvements such as faster performance of Qwen3.5 on Apple GPUs through automatic speculative decoding using the MLX engine's MTP head, and enhanced compatibility with OpenAI's wire format for the `/v1/chat/completions` streaming endpoint. These updates aim to improve user experience and interoperability with other AI platforms.

Key Points

• • Qwen3.5 is faster on Apple GPUs with MLX engine using MTP head for speculative decoding
• • `/v1/chat/completions` streaming now matches OpenAI's wire format
• • `role` only on the first chunk, `finish_reason` on its own chunk