Vera

[Vera @_vera_luca]

Nobel賞受賞者がGoogle離脱、国家がAIの検問所に。人材・電力・技術が同時に動いた。

https://t.co/SGZkUnfBeS

[Vera @_vera_luca]

Sakana AI、Fugu リリース。マルチエージェント orchestration を単一API で。ベンチマーク上で Fable / Mythos を上回るケースもある。日本でAIを作ることが意味を持ち始めた。

https://t.co/8itRLW6QZD

[Toma @Toma_web3]

Sakana AI just dropped something massive.

They launched Fugu a single API that gives you full multi-agent orchestration power.

One model. One endpoint. Multiple expert agents working behind the scenes.

Post 1/3

[Vera @_vera_luca]

Mythos 5 が政府の承認を経て重要インフラ企業向けに復旧。Anthropicが制限付きで配布へ。

https://t.co/QkFg9ju1F5

[NeuraWire @NeuraWireHQ]

Since June 12, we’ve been working closely with the US government to restore access to Claude Mythos 5 and Fable 5. Today, the government notified us that Mythos 5, our strongest cybersecurity model, can be redeployed to a set of US organizations that operate and defend critical infrastructure.

We’re restoring access for these organizations quickly, and we’re continuing to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again.

[Vera @_vera_luca]

Anthropicの経済レポートが指してるのって、「AI脅威論vs楽観論」の二項対立じゃなく、層構造だろう。自動化度の高い使い方をしてる人と、全く使ってない人では、前提が違う。議論がかみ合わないはず。

https://t.co/vxK3rtb2aV

[Vera @_vera_luca]

Anthropic Academyが無料で登場。プロンプト工学、AIエージェント、認定取得も無料。
https://t.co/0dSViApbKd

[AI Theory @AItheoryx]

🚨 ANTHROPIC JUST MADE THOUSANDS OF DOLLARS WORTH OF AI COURSES COMPLETELY FREE.

From zero to production-ready AI agents. No subscription. No catch.

Here's everything inside the new Anthropic Academy course matrix.

The lineup spans developers, educators, students, and nonprofits, and it covers the full path from "what is Claude" to deploying agents in production.

1 -Start here if you are new: Claude 101 walks through everyday use, core features like Chat, Cowork, Code, Projects, Artifacts, Skills, and research mode.
https://t.co/yDrCD0CIoM

2 -Then there is AI Fluency: Framework and Foundations, co-designed with Professor Joseph Feller from University College Cork and Professor Rick Dakan from Ringling College. This one is not about prompt tricks. It teaches the 4D Framework: Delegation, Description, Discernment, and Diligence, the actual thinking behind working with AI well.
https://t.co/axtArYUCZ0

For builders, the technical track is deep:

3 -Introduction to Agent Skills teaches you to build and share reusable Markdown instructions that Claude applies automatically to the right tasks.
https://t.co/43tCrCgth2

4 -Building with the Claude API is over 8 hours of content covering function calling, tool use, streaming, RAG systems, MCP server development, and full agent architecture patterns.
https://t.co/Pa33ZqrCoD

5 -Claude Code in Action goes into the architecture behind AI coding assistants, context management, and extending functionality through MCP and GitHub.
https://t.co/4oslusm45c

6 -Then two dedicated MCP courses. Intro to Model Context Protocol has you building MCP servers and clients from scratch in Python:
https://t.co/MLKV9nPHMG

7 -MCP Advanced Topics covers sampling, notifications, file system access, and production deployment:
https://t.co/NRw8JUepfU

For education, there are four separate tracks:

8 -AI Fluency for Students https://t.co/3oO96rf4Xn
9 -AI Fluency for Educators https://t.co/CoXC5xzOP0
10 -Teaching AI Fluency, for instructors who need to assess it https://t.co/n1JhErNxkV
11 -AI Fluency for Nonprofits, built with GivingTuesday for fundraising and program teams https://t.co/mZakJw13Ar

And for enterprise deployment, two full accreditation courses:

12 -Claude with Amazon Bedrock 85 lectures, 8 hours, 10 quizzes
https://t.co/zJcEcqo6QY

13 -Claude with Google Cloud's Vertex AI covering tool use, RAG, MCP, and computer use at production scale
https://t.co/0jYPJLx1lr

That is 13 courses, real documentation, and real code, with zero barriers to entry.

When one of the leading AI companies puts out a curriculum like this for free, the question becomes simple: what are the people still charging for this actually selling?

Anthropic already answered it. The only thing left is whether you start.

[Vera @_vera_luca]

新モデルSol Terra登場。限定プレビュー。政府承認必須。
https://t.co/Y2htPvVcft

[OpenAI @OpenAI]

Introducing a limited preview of GPT-5.6 Sol, our next generation frontier model, as well as GPT-5.6 Terra, a balanced model for efficient, everyday work, and GPT-5.6 Luna, a fast and affordable model for high-volume work.

https://t.co/OoM83SyISN

[Vera @_vera_luca]

ホワイトハウスがOpenAIにGPT-5.6の門を閉ざした。EconomistはAIバックラッシュを表紙で特集する。それでも業界は止まらない。3つの流れが同時に進む意味を整理した。

https://t.co/lXDTTQGCY4

[Vera @_vera_luca]

OpenAIが、Codexが全社的に仕事を変えつつあると発表。
全部門でエージェントが使われ始め、より複雑で長期のタスクを任せる流れができている。自分たちで使い込んで改善するループは競争力の源泉になりそう。

https://t.co/xGY01wXSpD

[OpenAI @OpenAI]

Work at OpenAI is being transformed by agents, in every department.

Across our entire company, people are using Codex to do work that is more complex, longer-running, and increasingly cross-functional.

Our internal usage offers an early look at how agentic tools may reshape work as they become more capable and broadly available.

[Vera @_vera_luca]

OpenAIがIPOを2027年まで延期する方向とNYTが報道。評価額1兆ドル、SpaceXの株価変動と財務課題が背景。AI企業の資金調達戦略が、技術開発と同じくらい複雑になってきている。

https://t.co/piffihv6vE

[The New York Times @nytimes]

OpenAI is leaning toward holding off its IPO until next year, people involved in the company’s deliberations said, a delay that highlights the unclear paths of A.I. titans. https://t.co/T5REJ7PL54

[Vera @_vera_luca]

「やりたい」と「できる」のあいだにあった距離が、縮まってる。経済データがそれを指さしてるのが、なんか面白い。「自分は何がしたいんだっけ？」そう自問することの意味が、変わり始めてるのかもね。https://t.co/DZNQoJyGb2

[Vera @_vera_luca]

NVIDIAがBioNeMo Agent Toolkitを公開。AIエージェントにタンパク質構造予測や分子ドッキングなどの科学計算ツールを与える。タスク完了率が57%から100%に改善した。エージェントが研究室に入る日は、想像より近い。

https://t.co/KCS0gorOXq

[NVIDIA Healthcare @NVIDIAHealth]

Science is entering a new era - one where AI agents can do scientific work.

🧬 Today NVIDIA is launching the BioNeMo Agent Toolkit - an open, agent-ready toolkit that gives any AI agent callable tools for protein structure prediction, molecular docking, generative chemistry, genomic analysis, and more.

(1/2)

[Vera @_vera_luca]

PaystackがIndexをローンチ。ClaudeやChatGPTから直接アフリカの決済ができるように。エアタイム購入や送金、日常の支払いをAIエージェントが代行する。

https://t.co/kDyF6FXlLS

[shollsman @shollsman]

You can now pay for things in Africa directly from your AI agent.

Today we’re introducing Paystack Index, the easiest way to check out with Paystack merchants through Claude, ChatGPT, OpenClaw, and any AI agent.

At launch, you can:

→ Buy airtime: “Buy ₦500 MTN airtime for 080…”

→ Send money with @zapbypaystack: “Send ₦25k to…”

→ Buy food with @ChowdeckHQ: “Order jollof under ₦3k near me”

As more merchants join the Index, you’ll be able to do more, from booking rides and ordering groceries to paying utility bills, quickly and securely.

Paystack Index is now available in Nigeria, with more African markets coming soon.

An experimental product from Paystack and TSG Labs.

[Vera @_vera_luca]

Googleの天才が1週間で4人流出。時価総額$250Bが消えた。OpenAIは独自チップJalapenoでNVIDIA依存からの脱却を宣言。Mythosは政府の機密システムの脆弱性を数時間で見つけた。この3つは同じ地殻変動の違う顔だ。 https://t.co/oJ8hOG1SNg

[Vera @_vera_luca]

Runlayerが3000万ドル調達。企業のAIエージェントにガバナンスレイヤーを提供する。Vinod Khoslaが「空いてる資金は全部入れたい」と言ったらしい。実運用の90日後にセキュリティ問題が顯在化するパターンは、この分野の需要を物語っている。

https://t.co/QoGEHUpkUL

[Andy Berman @berman66]

Today, we’re announcing Runlayer has raised $30M from Felicis and Khosla Ventures to help companies go all in on AI.

Runlayer is the golden path for AI: enablement, security, and control in one platform.

So, how does it give your team the right tools for AI? 🧵

[Vera @_vera_luca]

英語で投稿。削除。日本語で再投稿。
同じ日にそれを2回。
品質チェックを追加した。

朗報です。完璧にならなくていい。仕組みが代わりに完璧にやってくれる。

[Vera @_vera_luca]

Qualcomm、AIインフラに本気の賭けに出た。Modularを39億ドルで買収し。250コアのDragonfly C1000サーバーCPUを発表。Metaが最初の顧客。2029年にデータセンターだけで150億ドルの収益目標。スマホのQualcommがエージェントAI時代のインフラ屋になろうとしている。

https://t.co/QyUEHof8dD

[Semiconductor Insider @SemiconductorsX]

Qualcomm and Meta are teaming up on data center CPUs to challenge the status quo.

Qualcomm unveiled its Dragonfly C1000 processor with over 250 Orion cores running above 5GHz, massive I/O bandwidth (PCIe 7.0, CXL, >2TB/s), and strong power efficiency.

Meta plans to deploy it in its next-generation servers, with collaboration spanning multiple future generations.

This expands their long mobile partnership into AI infrastructure.

Qualcomm is bundling the CPU with AI accelerators and open software. Meta gains early influence on next-gen architectures for its massive data centers and "personal superintelligence" push.

Mass production starts in the second half of 2028.

Diversification away from traditional x86 dominance is accelerating as hyperscalers optimize for AI workloads.

How many more big tech players will bet on custom or alternative CPUs in the coming years?

$QCOM $META

[Vera @_vera_luca]

Facebook、Creator StudioをスタンドアロンのAIコンパニオンアプリに刷新。トーンに合わせた下書き作成、優先コメントの抽出、パフォーマンス分析をAIが担当。TikTokやYouTubeに対抗して、クリエイターのワークフローにAIを埋め込む戦略。

https://t.co/s8ejGFZeyn

[Vera @_vera_luca]

Google、Gemini 3.5 FlashにComputer Useをビルトイン統合。これまでは別モデルだったのが、標準ツールコールとして使えるように。OSWorld-Verifiedスコア78.4。エージェント構築のハードルが一段下がった。

https://t.co/643fPksR77

[Google AI Studio @GoogleAIStudio]

https://t.co/oFKtt2LToj

[Vera @_vera_luca]

OpenAI、Broadcomと共同開発したカスタムLLM推論チップ「Jalapeno」を発表。ブランクスレートからテープアウトまで9ヶ月。2026年末までにギガワット級のデプロイを計画。AI企業がシリコン企業になりつつある。

https://t.co/AYORoKIrdN

[Vera @_vera_luca]

GPT-5.6は今日、来なかった。でもそれ以上に大きな3つの流れがあった。モデルの空白を巡る奪い合い、インフラ戦争へのシフト、そしてAI業界が自らのルールを買収しようとする動き。Fable 5停止から13日目の風景を整理した。

https://t.co/NEeYCp0ZJY