PulseAugur
实时 12:53:51
English(EN) 🚀 Ollama v0.32.4 is out What's Changed - Support Laguna on Apple GPUs via the MLX engine - Quantize draft-model output heads at the requested type when creating

Ollama v0.32.4 通过 Apple MLX 支持增强本地 AI,并支持多模态视觉

Ollama 发布了 v0.32.4 版本,为本地 AI 推理带来了显著的增强,特别是对于拥有 Apple Silicon 硬件的用户。此次更新通过 Apple 的 MLX 引擎支持 Laguna 模型系列,从而在集成 GPU 上实现加速推理。此外,该版本通过量化 draft-model output heads 来优化投机解码,以提高效率和准确性,并修复了 Qwen3 MoE 模型的解码问题。此次更新还包括 llama.cppMinimax-M3 模型的初步视觉支持,扩展了其多模态能力。 AI

影响 增强了 Apple Silicon 用户本地 AI 推理能力,并扩展了 llama.cpp 中的多模态支持。

排序理由 这是一个用于促进本地 AI 模型推理的工具的软件发布,而不是前沿模型发布或核心研究。

在 Mastodon — fosstodon.org 阅读 →

AI 生成摘要 · Google Gemini · 来自 6 个来源。 我们如何撰写摘要 →

Ollama v0.32.4 通过 Apple MLX 支持增强本地 AI,并支持多模态视觉

本文如何被排名

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
这是一个用于促进本地 AI 模型推理的工具的软件发布,而不是前沿模型发布或核心研究。
Source corroboration
6 independent sources
Strong cross-source corroboration — multiple independent publishers covered this within the clustering window.
Topics
product, infra
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
59 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.
Coverage growth since scoring
+2 source(s) since last score
New sources have picked up this story since our last re-score. Score will update on the next scoring pass.

完整方法见我们的编辑标准

报道来源 [6]

  1. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    🚀 Ollama v0.32.5 发布了 有什么新变化 · * 修复了一个 MLX Metal 错误,该错误可能导致 NVFP4 模型(尤其是 Laguna)的输出质量下降。 · 完整更新日志:… ⭐ 1

    🚀 Ollama v0.32.5 is out What's Changed · * Fixed an MLX Metal bug that could reduce output quality for NVFP4 models, particularly Laguna. · Full Changelog:… ⭐ 177,021 stars All open-source AI releases → https:// opensourceai.tech/releases.html # OpenSource # Release # DevTools # …

  2. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    ⚙️ Ollama 新版本发布!⚙️ 版本:v0.32.5 发布说明:## 变更内容 * 修复了 MLX Metal 错误,该错误可能导致 NVFP4 模型的输出质量下降,特别是

    ⚙️ New Ollama Release! ⚙️ Version: v0.32.5 Release Notes: ## What's Changed * Fixed an MLX Metal bug that could reduce output quality for NVFP4 models, particularly Laguna. **Full Changelog**: https:// github.com/ollama/ollama/compa re/v0.32.4...v0.32.5 🔗 https:// github.com/olla…

  3. dev.to — LLM tag TIER_1 English(EN) · soy ·

    Ollama v0.32.4 支持 Apple MLX — 并带来 Rubin、llama.cpp 及 GPU 更新

    <p>Today's digest highlights Ollama v0.32.4 with new Apple MLX GPU support, alongside llama.cpp gaining preliminary vision capabilities. Additionally, we see releases for AMD ROCm 7.14.0 and TensorRT-LLM, plus details on NVIDIA's Rubin GPU architecture.</p> <h2> Local AI &amp; Op…

  4. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    🚀 Ollama v0.32.4 发布 - MLX 引擎通过 Apple GPU 支持 Laguna - 创建时量化草稿模型的输出头到请求的类型

    🚀 Ollama v0.32.4 is out What's Changed - Support Laguna on Apple GPUs via the MLX engine - Quantize draft-model output heads at the requested type when creating… ⭐ 176,886 stars All open-source AI releases → https:// opensourceai.tech/releases.html # OpenSource # Release # DevToo…

  5. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    ⚙️ Ollama 新版本发布!⚙️ 版本:v0.32.4 发行说明:## 变更内容 - 通过 MLX 引擎支持 Apple GPU 上的 Laguna - 量化 draft-model 输出头

    ⚙️ New Ollama Release! ⚙️ Version: v0.32.4 Release Notes: ## What's Changed - Support Laguna on Apple GPUs via the MLX engine - Quantize draft-model output heads at the requested type when creating speculative-decoding drafts. - Fixed Qwen3 MoE decoding for differently-quantized …

  6. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    ⚙️ Ollama 新版本发布!⚙️ 版本:v0.32.3 发行说明:## 变更内容 - 修复了在发送数据前卡住的模型下载问题。 - 改进了集成:res

    ⚙️ New Ollama Release! ⚙️ Version: v0.32.3 Release Notes: ## What's Changed - Fixed model downloads that stall before sending data. - Improved integrations: restored Claude Code Channels, fixed Anthropic thinking streams, and made Hermes Desktop respect `--force-build`. - Expande…