FeiZ

Personal Agent Dev · Economics·AI Engineer
Specializing in how AI can be applied to life
Let's connect!!

[FeiZ @Fei2411]

竟然真的有很多人不知道啊，OpenAI官方有验证GPT 生成内容的网站

很多人天天在那发各种照片，你往这一丢就知道了，几乎全是AI的

https://t.co/pra6RYprko

[FeiZ @Fei2411]

@wangchangfu88 反差？

[FeiZ @Fei2411]

视频生成模型是目前利润最大、最赚钱的模型领域

在这种情况下他们仍然将H3开源，我相信这证明了他们实现AGI的决心

[MiniMax (official) @MiniMax_AI]

The road to AGI was never meant to be traveled alone.

From Hailuo 01 to H3, we've been incredibly lucky to build alongside so many friends, partners, and fellow builders who believed in the same thing we did.

That's why we open-sourced H3. Because progress moves faster when more people can build on it, and because we want to keep moving forward together.

"To the edge of the sky" has always been more than a destination for us. It's a belief, and maybe a kind of idealism.

To everyone who has been part of this journey: thank you.

Together, toward the edge of the sky, we will keep going until we get there.🌌✨

[FeiZ @Fei2411]

@dandreyang @tianyi 谢谢！

[FeiZ @Fei2411]

多少上下文？10M ？
单个GPU？？

[Pokee AI @Pokee_AI]

Releasing Pokee-Isaac 28B — the world’s first real 10M-token context frontier-class agentic model, deployable on a single GPU (starting from RTX 4090 or equivalent).

New proprietary non-decoder-only architecture:
• 93.3% RULER at 10M tokens
• Up to 137K tokens/s prefill on one B200 with 10M-token context
• Leads BFCL v4 and τ³-bench in our evaluation
• Lowest combined attack success rate among evaluated models on DTAP security red-teaming benchmark

Pricing and deployment:
💰 $0.15/M input · $1/M output
🔒 Deploy in your VPC, on-premises, or on-device, with Day-0 support for @vllm_project and @sgl_project

Technical blog: https://t.co/nFqaYBlcQP
Technical report: https://t.co/XDOoZxpgJx
API: https://t.co/KYj8fOOjNS

[FeiZ @Fei2411]

据悉：特朗普政府的自愿性前沿AI安全框架主要针对闭源专有模型，开放权重模型被豁免发布前测试。

[FeiZ @Fei2411]

黄拉起的开放模型联盟还是很有成果的

特朗普政府2026年6月2日签署行政令，要求建立一套评估AI模型高级网络安全能力的机制，并设计自愿框架：符合条件的“覆盖前沿模型”在正式发布前，可向联邦政府提供最多30天访问窗口，用于安全测试与协作。行政令明确强调，这不构成任何强制性许可或预审要求。

近期泄露的框架实施细节明确：

1.自愿测试主要针对达到前沿网络安全与黑客能力基准的闭源专有模型。
2.开放权重模型（open-weight models）被豁免，不纳入这套发布前政府测试要求。
3.框架重点放在美国公司开发的闭源前沿模型上，开放模型（包括美国公司出品的开放权重模型）不在强制或默认测试范围内。

整体定位是自愿参与、聚焦网络安全风险，同时避免对开放模型生态施加额外发布限制。

[FeiZ @Fei2411]

黄拉起的开放模型联盟还是很有成果的

特朗普政府2026年6月2日签署行政令，要求建立一套评估AI模型高级网络安全能力的机制，并设计自愿框架：符合条件的“覆盖前沿模型”在正式发布前，可向联邦政府提供最多30天访问窗口，用于安全测试与协作。行政令明确强调，这不构成任何强制性许可或预审要求。

近期泄露的框架实施细节明确：

1.自愿测试主要针对达到前沿网络安全与黑客能力基准的闭源专有模型。
2.开放权重模型（open-weight models）被豁免，不纳入这套发布前政府测试要求。
3.框架重点放在美国公司开发的闭源前沿模型上，开放模型（包括美国公司出品的开放权重模型）不在强制或默认测试范围内。

整体定位是自愿参与、聚焦网络安全风险，同时避免对开放模型生态施加额外发布限制。

[MTS @MTSlive]

SITUATION UPDATE: Only open models made by US companies will be exempt from safety testing under the new guidelines, per WSJ.

[FeiZ @Fei2411]

@PopBase Wait, is this actually real? Unbelievable.

[FeiZ @Fei2411]

@PopBase My wallet is absolutely not ready for this.

[FeiZ @Fei2411]

埃隆马斯克最新电话会议透露：
Grok 4.6 大概下周上线，Grok 4.7 大约3-4周后推出。AI 迭代节奏接下来会明显加快。

Grok 5 预计今年年底前发布，会把 SpaceX 过去25年积累的全部数据全部喂进去训练。埃隆说，这会让 Grok 成为「目前为止最好的工程师」。

[FeiZ @Fei2411]

@PopBase This is wild. Let's see how it plays out.

[FeiZ @Fei2411]

@dokagui_anime これは神アプデの予感。

[FeiZ @Fei2411]

@MLavca クオリティ高すぎてしんどい。

[FeiZ @Fei2411]

@flurry 完全に予想外の展開で笑う。

[FeiZ @Fei2411]

@tncoin6 哈哈哈，行

[FeiZ @Fei2411]

刚在地铁上想到个好点子，但因为正好要下车了所以没有及时记录下来

我现在气的想扇自己一巴掌

[FeiZ @Fei2411]

@Seven_Nguyen666 Đúng vậy, khi bạn tạo ra giá trị thật thì mạng xã hội cũng có thể trở thành một công việc.

[FeiZ @Fei2411]

@hoshusokuhou こういうのでいいんだよ、こういうので。

[FeiZ @Fei2411]

@ngnchiikawa センスの块すぎて大好き。

[FeiZ @Fei2411]

@beni_yakumo マジか、これはアツすぎる。