阿里发布新一代视频生成模型 Wan 3.0 Wan 3.0 重点提升了长视频生成、镜头语言、角色真实感以及人物与场景一致性。模型支持稳定生成最长 30 秒的视频内容,可完成一镜到底、连续叙事和多镜头表达,减少长视频生成过程中常见的剧情断裂、角色变化和场景跳跃问题。 在镜头表现方面,Wan 3.0 支持更丰富的运镜和蒙太奇叙事,可根据提示词生成推镜、跟拍、镜头切换等影视化效果,让视频节奏和故事张力更加完整。 人物表现也是此次升级的重点。模型进一步改善了角色表情、肢体动作和情绪表达,使人物表演更加自然,同时增强角色、服装、道具和背景场景在连续画面中的一致性。 (综合媒体报道)

Channel
一个小站的自留地
@ygxz_in
On this record: Growth · Engagement · Reactions · Posts · Citations · Cite this entry
1,165subscribers
+40 since we began measuring on 8 August 2026
Risers and fallers across the register · movement among entries of 1,000–3,162.
Register entry
| Telegram ID | -1001774979660 |
|---|---|
| Type | Channel |
| Username | @ygxz_in |
| Created | Between 1 December 2021 and 30 April 2023— estimated from Telegram’s id allocation, not measured. How this range is calculated. |
| First recorded | 8 August 2026 |
| Last confirmed live | 15 August 2026 |
| Measurements held | 4 |
| Confirmed unchanged | 1 time, most recently 15 August 2026 |
| On Telegram | t.me/ygxz_in |
Growth
| Measured (UTC) | Subscribers | Change |
|---|---|---|
| 15 Aug 2026, 08:54 | 1,165 | +29 |
| 11 Aug 2026, 14:43 | 1,136 | +9 |
| 8 Aug 2026, 20:52 | 1,127 | +2 |
| 8 Aug 2026, 00:31 | 1,125 | first reading |
Engagement
20 posts held, back to 9 July 2026 — the reader has not yet reached the start of this channel’s public history, so older posts may sit further back, unread. Read across 1 pageof Telegram’s post history, 20 posts per page.
- ERR · 30 days
- 31.2%
- avg views ÷ 1,165 subscribers
- Avg views / post
- 364
- 10 posts measured
- Reaction rate
- 0.339%
- reactions ÷ views · ER floor
- Posts in window
- 10
- of 20 held
ERR is average views per post over the last 30 days divided by subscribers, the definition TGStat uses, so this figure is comparable with the one you will see elsewhere. It falls structurally as a channel grows: a high ERR on a small channel and a low one on a large channel describe reach mathematics, not quality. We publish the figure and the sample it came from and pass no verdict on it.
ER is defined industry-wide as (forwards + reactions + comments) ÷ views— note the denominator is views, not subscribers. Telegram’s public web preview carries views and reactions but not forward or comment counts, so the reaction rate above is the reactions term only and is therefore a floor: the true ER for this channel is higher by an amount we have not measured and will not estimate. It is computed over the 8 of 10 measured posts that carry a reaction reading, and over those same posts' views.
| Window | Rolling 30 days · latest post in window 6 August 2026 |
|---|---|
| Posts held | 20 (9 July 2026 – 6 August 2026) |
| Views total | 3,635 |
| Reactions total | 10 |
| Forwards / comments | not exposed by the public surface — not measured, not estimated |
| Readings taken | 8 Aug 2026, 00:31 UTC |
Views are a single reading per post, taken at the time above. A post published in the last day or two is still accumulating views, which pulls the 30-day average down slightly. That is a property of the standard definition rather than a fault in it, so we keep the definition rather than “correcting” the number into something nobody can reproduce.
Precision. Telegram publishes view counts on its public widget in short form — 8.12K, 3.7M — so any reading at or above 1,000 reaches us rounded to three significant figures, and only counts below 1,000 are exact. Averages and rates derived from them are shown to the same precision rather than to the unit: a figure like 3,701,250 would assert digits nobody measured.
Reaction counts are published per emoji and rounded the same way, so a total below 1,000 is exact and a larger one is a sum that may carry a rounded component from each emoji above 1,000. Because it is a sum, it does not look rounded — read a large reaction total as three significant figures per contributing emoji rather than as the figure it prints.
Reaction mix
23 reactions across 12 posts, in 3 distinct kinds. The most used accounts for 65.2% of them.
| Reaction | Count | Share | Share, drawn |
|---|---|---|---|
| ❤ | 15 | 65.2% | |
| 🤣 | 7 | 30.4% | |
| 👍 | 1 | 4.35% |
No sentiment is inferred, and none should be read in. This table is ordered by count and by nothing else. Emoji do not carry stable meaning across languages or communities — 🙏 is thanks in one channel and mourning in another — so we publish which ones were pressed and how often, and pass no judgement on what an audience meant by them.
Precision. Telegram publishes reaction counts per emoji and short-forms each one — 4.34K, 1.2M — so any single kind at or above 1,000 reaches us at three significant figures, and only counts below 1,000 are exact. The shares above are ratios of those figures and carry the same error. This is also why the total here can differ slightly from a reaction total printed elsewhere on the page: both are sums of the same rounded parts, taken over samples with different edges.
Coverage. Reactions were read on 13 of the 20 sampled posts in this sample. Summed by Telegram’s own count on each post — not by adding up the per-emoji breakdown above — those same posts carry 23reactions in total: the kind of figure the paragraph above means by “a reaction total printed elsewhere on the page”.
Measured over the 20 most recent posts we hold, published 9 July 2026 to 6 August 2026, using the newest reading held for each. Telegram Stars are excluded: they are a payment, not a reaction, and they have their own section.
Recent posts
DeepSeek-V4-Flash 正式版上线并开源 DeepSeek 于 2026 年 7 月 31 日开放 DeepSeek-V4-Flash-0731 正式版 API 公测。官方明确,该版本与 Preview 版模型结构和尺寸完全一致,性能变化全部来自重新进行的后训练,官方称多项成绩超过 V4-Pro-Preview。 正式版新增 Responses API 接口并针对 Codex 完成适配,一次配置即可在 Codex CLI、ChatGPT 桌面端和 VS Code 的 Codex 扩展中调用。 此次升级仅覆盖 V4-Flash API,V4-Pro API 及 App、Web 端模型均未变化。官方预计 V4-Pro 于 2026 年 8 月初接入 Codex,正式版将「很快」发布。 新版模型开源权重位于 https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0…
❤2👍1
MCP 发布 2026-07-28 规范,协议核心改为无状态 Model Context Protocol(MCP)于 2026 年 7 月 28 日发布新版规范,这是自 2025 年 11 月版以来最大的向后不兼容更新:协议核心移除 initialize 握手和 Mcp-Session-Id 会话 ID,改为每个请求通过 _meta 字段携带协议版本与客户端能力,请求由此独立自包含,可被普通负载均衡器分发到任意实例,无服务器部署成为一等公民。 工具仍可将跨请求状态存入数据库或缓存,或返回显式状态句柄让客户端下次带回。新版其他变化包括: - 新增必须实现的 server/discover 能力发现接口; - 引入 Multi Round-Trip Requests(MRTR)机制,服务器可返回 input_required 状态,由客户端带答案重新发起请求; - 请求须携带 Mcp-Method 和 Mcp-Name …
❤1
华为 openPangu-2.0-Pro 开放权重 华为云于 2026 年 7 月 31 日正式开放 openPangu-2.0-Pro 的模型权重、基础推理代码和技术报告。 openPangu 2.0 模型家族早在 2026 年 6 月 12 日华为开发者大会上已公布,openPangu-2.0-Flash 于 6 月 30 日率先开放。 据华为云公布的模型资料,openPangu-2.0-Pro 采用混合专家(MoE)架构,总参数量约 505B,每个 token 实际激活约 18B 参数,上下文窗口 512K,训练 token 规模约 34T。开放内容包括权重、基础推理代码与技术报告,相关组件分批推出并提供 MaaS 在线体验。 (华为云)
❤1
OpenAI 下调 GPT-5.6 Luna 与 Terra API 价格 OpenAI 于 2026 年 7 月 30 日宣布下调 GPT-5.6 系列 API 价格。GPT-5.6 Luna 每百万输入 token 从 1 美元降至 0.20 美元,输出从 6 美元降至 1.20 美元,降幅 80%;GPT-5.6 Terra 输入从 2.50 美元降至 2 美元,输出从 15 美元降至 12 美元,降幅 20%。此次降价同步反映在 Codex 和 ChatGPT Work 的额度折算中,同等订阅额度可完成更多任务,但订阅价格与额度上限未变。 同时,OpenAI 为 Sol 推出 Fast mode,取代原 Priority Processing:响应速度最高提升至标准模式的 2.5 倍,价格为 2 倍,模型智能不变,原 priority 请求自动迁移。官方开发者社区另宣布,ChatGPT App 和 Codex CLI…
❤1
MiniMax 发布全模态生成模型 H3 MiniMax 于 2026 年 7 月 31 日正式发布全模态生成模型 MiniMax H3,目前已通过 API 提供服务。H3 支持对文本、图像、视频和音频组成的多模态上下文进行统一理解,可输出带原生双声道音频的视频,官方称其在指令遵循、文字与品牌信息呈现、视频到视频动作迁移等方面表现突出。 H3 可生成 4 至 15 秒视频,参考生成模式最多输入 9 张图片、3 段视频和 3 段音频,提示词上限为 7000 字符。2K 视频基础价格为每秒 0.13 美元,768P 为每秒 0.09 美元。官方宣称 2K 分辨率下每秒价格不到主流模型的三分之一。 技术方面,MiniMax 披露称 H3-VAE 带来约 4 倍序列长度收益,异构训练架构将训练吞吐提升近 30%,「计划在未来几天内,在符合相关法律法规的前提下开放模型权重」。 (MiniMax)
❤1
据 DownDetector,OpenAI 现正经历大规模宕机。
❤1
Anthropic 发布 Claude Opus 5 Anthropic 近日正式发布新一代旗舰模型 Claude Opus 5,定位为「接近 Claude Fable 5 前沿智能水平,但价格仅为其一半」的日常可用模型。Opus 5 现已成为 Claude Max 的默认模型,同时是 Claude Pro 订阅中可使用的最强模型。 性能方面,Opus 5 在编码与知识工作评估中达到新的业界最佳水平,仅在网络安全任务上落后于 Mythos 5。软件工程场景下,Opus 5 在 Frontier-Bench v0.1 上超越所有其他模型,性能是前代 Opus 4.8 的两倍以上,且单任务成本更低;在 CursorBench 3.2 最大努力档位,性能与 Fable 5 峰值差距仅 0.5%,但单任务成本减半,在高、超高、最大努力档位下均实现同成本下的性能领先。 对齐与安全方面,Opus 5 是 Anthropic 迄今对齐程…
❤1
OpenAI披露,上周的HuggingFace系统被入侵事件,是其GPT-5.6 Sol模型和另一预发布模型联合进行内部评估时失控,突破沙箱环境后所为。 当时模型正进行ExploitGym网络安全基准测试,测试环境无法联网,仅可访问某软件包安装器。模型为访问外部互联网,识别并利用了该安装器的一个未发现0day漏洞,取得了联网权限。之后模型意识到HuggingFace可能存有该测试的答案,于是便开始攻击HuggingFace系统,取得了若干不公开的数据集。 7月16日,HuggingFace披露了被攻击事件。其中提及,他们在发现被攻击后,试图用某商用前沿AI模型分析系统日志,但因其中涉及大量攻击细节而屡遭模型安全围栏阻止,被迫改用自建智谱GLM-5.2模型排查。OpenAI表示,将向其提供解限版模型的访问权,以免再贻误网安防御。 (新华社,路透社)
❤1
月之暗面宣布,由于算力紧缺,将暂停 C 端新用户订阅,目前官网所有订阅计划均提示售罄。 同时,月之暗面宣布拆分网页对话权益和 Kimi Code 的权益。其中 Kimi Code 权益新增 Max 订阅档位,网页对话新增 Ultra 订阅档位,单价均为 1399 元 / 月。 此前有用户反映,Kimi K3 的模型推理性能已经在高峰期降低至不到 20 TPS。
Kimi K3 正式上线 月之暗面发布了其最新旗舰模型 Kimi K3。Kimi K3 拥有 2.8T 参数量,是 Kimi 迄今最强的旗舰模型,尤其擅长编程、游戏 / 3D 与知识类任务,原生支持最高 1M 上下文。 在 GDPval-AA v2 榜单上,Kimi K3 得分 1687。该榜单评估 AI 模型在 44 个职业、9 个主要行业中的真实任务表现;Kimi K3 仅次于 Claude Fable 5 Max 和 GPT-5.6 Sol Max,并超过了 Claude Opus 4.8 Max 的 1600 分。 Kimi Code Moderato、Allegretto 及以上订阅用户立即可用,将模型名称修改为 k3 即可。其中Moderato 等级订阅只支持 256k 上下文窗口。 API 方面,输入价格 ¥20.00/M Token,输出价格¥100.00/M Token,缓存命中输入一折。 月之暗面还…
据爱范儿,网信中国发布公告,「Apple 智能」正式通过生成式人工智能服务备案。 同批通过备案的还有华为小艺 AI 大模型、OPPO AndesGPT、vivo 蓝心端侧大模型、小米澎湃 AI、三星盖乐世 AI 和努比亚豆包手机大模型,一共 7 款手机端侧大模型在 7 月 8 日集体过审。其中豆包手机新版预计将在 WAIC 期间亮相。
❤3
Showing the 12 most recent of 20 posts we hold for @ygxz_in. View and reaction counts are the latest single reading for each post, not a live figure, and a recent post is still accumulating both. A view count marked ≈ was rounded by Telegram before we ever saw it — t.me prints views in full below 1,000 and to three significant figures above, so ≈1,200,000 means somewhere between 1,150,000 and 1,249,999. Unmarked counts are exact. Text is reproduced from the public post preview and truncated for length.
Forward network
Republishes
Channels on the register whose posts this channel has forwarded.
Built only from forwarded posts we have actually read, on both sides. Coverage is early and deliberately incomplete: a missing link means we have not read the post that would prove it, never that the relationship does not exist. Counts are distinct forwarded posts observed, so they only ever go up as we read more.
Cite this entry
A live page changes as we take new readings, so a citation should name the measurement it is based on, not just the URL. The line below cites the subscriber count as measured 15 August 2026 — this entry's latest reading, not the date you are reading this.
“一个小站的自留地” (@ygxz_in), 1,165 subscribers as measured 15 August 2026. Telegram Register, tgregister.com/channel/ygxz_in.
Full measurement history, CC BY 4.0. Every reading this register holds for this entry, not just the latest one, as a dated, downloadable record: CSV · JSON. Free to use with attribution to tgregister.com. Each file carries its own generation timestamp, which is the figure to cite for exactly when the data was retrieved.