📖 2 min read
Good evening. Quick rundown of what hit the AI feed in the last few hours — minus the noise.
1. GPT-5.5 lands on AWS Bedrock and Databricks
OpenAI’s GPT-5.5 is now generally available on Amazon Bedrock alongside Codex and Bedrock Managed Agents, and it dropped on Databricks the same day, governed through Unity AI Gateway. Translation: enterprises that live in AWS or Databricks can now wire frontier OpenAI models into their stack without leaving their existing security and observability fences. Big deal for procurement teams that have been stuck in pilot purgatory.
📧 Want more like this? Get our free The Ultimate AI Tool Database: 200+ Tools Rated & Ranked — Downloaded 5,000+ times
2. Pentagon signs classified AI deals with seven companies
The Defense Department inked contracts with seven AI firms — including OpenAI and Google — for classified military work. The notable detail: Anthropic reportedly rejected a “lawful use” clause in its own DoD contract, prolonging a public spat that started last month. Worth watching how the safety-positioning crowd squares this with growing defense revenue.
3. The Atlantic: maybe AI isn’t a bubble
The Atlantic’s new piece argues Anthropic’s revenue is climbing faster than Zoom in the pandemic, Google in the early 2000s, even Standard Oil in the Gilded Age. The driver, in two words: Claude Code. If you’ve been on the “this is all vibes” side of the debate, this one’s worth a read — the unit economics are starting to actually show up.
4. Moonshot ships FlashKDA
Moonshot AI released FlashKDA, a high-performance implementation of Kimi Delta Attention that plugs directly into the flash-linear-attention ecosystem. Early benchmarks show meaningful speedups. Open-source attention research keeps eating efficiency wins that were “frontier lab only” 18 months ago.
📧 Want more like this? Get our free The Ultimate AI Tool Database: 200+ Tools Rated & Ranked — Downloaded 5,000+ times
🔥 Hot take of the day
The community verdict on GPT-5.5 is settling: “For general chat, summarization, or creative writing, the improvements over 5.4 are marginal. For agentic work — automated pipelines, coding agents, long-horizon tasks — it’s a meaningful upgrade.” Translation for everyone else: if you’re not building agents, you can probably skip the upgrade hype. If you are, it’s actually moving the needle.
Quick links
- Tracking which AI tools are actually worth paying for? AiToolCrush rounds them up.
- Curious whether AI moves prediction markets the way it moves stocks? BetOnAI tracks the overlap.
That’s the wrap. Catch you in the morning.