PulseAugur
中
实时 18:53:33
English(EN) Roughly 57 percent on Terminal-Bench 4.1 for Gemini 4 Hydrogen, vs 18.2 percent for Gemini 3.8 Flash. Rolling out via AI Studio and the Gemini API, replacing Fl

Google 的 Gemini 4 Hydrogen 在基准测试中击败 Flash 模型,通过 API 推出

Google 发布了 Gemini 4 Hydrogen,在 Terminal-Bench 4.1 基准测试中取得了 57% 的得分,显著优于 Gemini 3.8 Flash 的 18.2% 得分。这款新模型可通过 AI Studio 和 Gemini API 获取,旨在取代需要速度的任务(如编码和代理工作)的中端模型。定价结构为超过一百万个 token 提供统一费率,旨在为高吞吐量的代理应用提供成本可预测性,而 Carbon 和 Argon 模型仍可用于需要深度推理的任务。 AI

影响 为中端模型设定了新的性能标准,有可能通过可预测的定价加速代理应用。

排序理由 前沿实验室模型发布,包含基准测试得分和 API 可用性。[lever_c_demoted from frontier_release: ic=1 ai=1.0]

在 Mastodon — mastodon.social 阅读 →

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

Google 的 Gemini 4 Hydrogen 在基准测试中击败 Flash 模型,通过 API 推出

本文如何被排名

Signal score
17 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Significant
前沿实验室模型发布,包含基准测试得分和 API 可用性。[lever_c_demoted from frontier_release: 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 — mastodon.social TIER_1 English(EN) · [email protected] ·

    Gemini 4 Hydrogen 在 Terminal-Bench 4.1 上得分约 57%,而 Gemini 3.8 Flash 得分为 18.2%。通过 AI Studio 和 Gemini API 推出,取代 Fl

    Roughly 57 percent on Terminal-Bench 4.1 for Gemini 4 Hydrogen, vs 18.2 percent for Gemini 3.8 Flash. Rolling out via AI Studio and the Gemini API, replacing Flash for mid-tier models for speed in coding and agentic work. Carbon and Argon stay for deep reasoning. Flat pricing ove…