Introducing Claude Opus 5.5, the first model in our new Claude 5.5 family.
It performs at the level of Claude Fable 5.1 for most tasks, and costs 40% less to run than Opus 5.
Opus 5 is a great model for coding, data analysis, design, biology, knowledge work.
More than any of these eval scores, what is most exciting to me is something else: Opus 5 is our least prompt injectable model yet. It is a bit buried in the system card, but across PI evals and red teaming, Opus 5 is very hard to prompt inject successfully.
And when layering defenses -- strong model alignment, combined with prompt injection probes, combined with Auto Mode in Claude Code -- the success rate for prompt injection attacks drops to ~0. This is new and exciting! More about this soon.
https://t.co/Tc7z2FqJhQ
Preview Claude Design cho 3D object
Ưu điểm: hỗ trợ export sang các file cho dân l��m 3D như GLB, OBJ.
Hạn chế: Không render được các vật thể đòi hỏi độ chi tiết cao.
Bài test thử nghiệm trên option cao nhất hiện tại: model Fable 5, effort max.
Higgsfield MCP now connects to Claude! 🧩
The first way to generate visuals on Claude, powered by Seedance 2.0, GPT Images 2.0, Marketing Studio and Cinema Studio.
Research on Claude. Polish your prompts. Generate ads, videos and brand content via the Higgsfield connector.
What's new in GPT-5.5:
- Stronger coding + debugging
- Better computer use (operating software autonomously)
- Deep research + doc/spreadsheet generation
- Greg Brockman: "It can look at an unclear problem and figure out just what needs to happen next"
Rolling out now to Plus, Pro, Business, Enterprise in ChatGPT + Codex.
The AI model war is compressing release cycles down to weeks. Wild time to be building.
https://t.co/aXISpFx6Z1
GPT-5.5 just dropped. Less than 2 months after GPT-5.4.
OpenAI's release cadence is getting insane. Better coding, computer use, deep research with "less guidance needed."
Meanwhile Anthropic's Claude Mythos has Wall Street talking.
Is OpenAI shipping fast to stay relevant, or are they genuinely pulling ahead?
What happened (April 18, 2026):
- Kelp's LayerZero V2 bridge exploited
- 116,500 rsETH minted out of thin air
- Deposited into Aave V3 on ETH + Arbitrum as collateral
- ~$190M in wETH/wstETH borrowed and gone
- Aave now sitting on up to $230M bad debt
Root issue: Aave accepted bridged assets as collateral with no oracle safeguard for bridge integrity.
Biggest DeFi exploit of 2026 so far.
https://t.co/xhPpdwVNkY
The Kelp DAO bridge exploit just created $123M-$230M bad debt on Aave. $9B in outflows followed.
Attacker minted 116,500 rsETH, dumped it as collateral, borrowed ~$190M ETH. Aave didn't get hacked. It got used as the exit.
Is DeFi collateral risk still unsolved?
Introducing Claude Opus 4.7, our most capable Opus model yet.
It handles long-running tasks with more rigor, follows instructions more precisely, and verifies its own outputs before reporting back.
You can hand off your hardest work with less supervision.
Introducing Gemini 3.1 Flash TTS 🗣️, our latest text to speech model with scene direction, speaker level specificity, audio tags, more natural + expressive voices, and support for 70 different languages.
Available via our new audio playground in AI Studio and in the Gemini API!
/ultraplan is the move I have been waiting for. Plan on the web, execute on local CLI, kill the back and forth of pasting plans into terminal.
Web plus CLI hybrid is the real Claude Code unlock for 2026.
New in Claude Code: /ultraplan
Claude builds an implementation plan for you on the web. You can read it and edit it, then run the plan on the web or back in your terminal.
Available now in preview for all users with CC on the web enabled.