PulseAugur
中
实时 16:32:03
Italiano(IT) Flint è un modello MoE da 40 miliardi di parametri progettato per il coding locale. Attivando solo 2,5 miliardi di parametri per token, raggiunge i 125 token al

Flint MoE 模型实现本地编码每秒 125 个 token

Flint 是一个拥有 400 亿参数的混合专家(MoE)模型,专为本地编码任务设计。它在 MacBook Pro M4 Pro 上实现了每秒 125 个 token 的速度,每个 token 仅激活 25 亿参数,性能优于同等密度的模型。该模型由 Eureka Machine 在四个月内开发完成。 AI

影响 提供了一个更快速、本地运行的编码助手,可能改善开发人员的工作流程和隐私。

排序理由 发布了一个新的开源模型,并附带了性能指标。[lever_c_research降级:ic=1 ai=1.0]

在 Mastodon — fosstodon.org 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

Flint MoE 模型实现本地编码每秒 125 个 token

本文如何被排名

Signal score
8 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
发布了一个新的开源模型,并附带了性能指标。[lever_c_research降级:ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
model release, product
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
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

  1. Mastodon — fosstodon.org TIER_1 Italiano(IT) · [email protected] ·

    Flint is a 40B parameter MoE model designed for local coding. Activating only 2.5B parameters per token, it reaches 125 tokens/s

    Flint è un modello MoE da 40 miliardi di parametri progettato per il coding locale. Attivando solo 2,5 miliardi di parametri per token, raggiunge i 125 token al secondo su MacBook Pro M4 Pro, superando per velocità i modelli densi equivalenti. Sviluppato in meno di quattro mesi c…