@thenewstack

All about at-scale software development, deployment & management. Tech news, analysis, research, podcasts & more! Subscribe đź“© https://xitter.cf/t.co/X1jwJjLnNj

 We're Global 🌏
thenewstack.io
Joined February 2014
  • Tweets127K
  • Following3.1K
  • Followers41.4K
11.4KMedia
PHP’s security scramble is just a preview: AI is digging up flaws faster than maintainers can fix them thenewstack.io/php2shell-ai-…

PHP’s security scramble is just a preview: AI is digging up flaws faster than maintainers can fix...

PHP2Shell’s chaotic disclosure sparked a row. Behind the drama, maintainers warn that AI is piling new security work onto already stretched teams.

thenewstack.io
1
3
213
Five AI agent harnesses took the same memory test. The results weren't even close. thenewstack.io/letta-frontie…

Five AI agent harnesses took the same memory test. The results weren't even close.

Letta’s new FrontierMemory benchmark tests whether AI agents learn from experience, and the harness matters as much as the model behind it.

thenewstack.io
1
3
278
Anthropic built its own git replacement for AI-scale coding. But you can’t have it. thenewstack.io/anthropic-sci…

Anthropic built its own git replacement for AI-scale coding. But you can’t have it.

Claude helped Anthropic outgrow its Git infrastructure — and build the replacement. But the company says Scion is too specialized to release for other teams.

thenewstack.io
1
2
729
Nvidia squeezed a 74B agent model into 24GB so it can run on your laptop thenewstack.io/nvidia-nemotr…

Nvidia squeezed a 74B agent model into 24GB so it can run on your laptop

Nemotron 3.5 Bolt uses 2-bit quantization to fit a 74B mixture-of-experts model in 24GB, and Nvidia says it still beats Qwen and Gemma on agent benchmarks.

thenewstack.io
1
273
AI found a critical PHP flaw. The bigger failure began in 2024. thenewstack.io/php2shell-cha…

AI found a critical PHP flaw. The bigger failure began in 2024.

Two critical PHP flaws sent some sites offline. The fight over Chainguard’s early patch release may outlast the bugs themselves.

thenewstack.io
338
AI’s Midas problem: When code is free, features are worthless thenewstack.io/ai-commoditiz…

AI’s Midas problem: When code is free, features are worthless

If anyone with a Claude subscription can build it, it isn’t worth much. The value has moved to understanding what users actually need.

thenewstack.io
1
1
383
DeepSeek open-sources the inference engine behind its API. And Huawei gets equal billing with Nvidia. thenewstack.io/deepseek-deep…

DeepSeek open-sources the inference engine behind its API. And Huawei gets equal billing with...

DeepInfer lets developers self-host V4.1 Flash with DeepSeek’s own optimizations, but stops short of a complete production stack.

thenewstack.io
354
DeepSeek V4.1-Flash-1017 vs. Claude Haiku 5.5: More accurate at a fraction of the price thenewstack.io/deepseek-flas…

DeepSeek V4.1-Flash-1017 vs. Claude Haiku 5.5: More accurate at a fraction of the price

Haiku 5.5 missed just four of 1,600 fields. Two came from a date it shouldn’t have trusted

thenewstack.io
386
Google turns the infrastructure behind Gemini’s agents into a cloud service thenewstack.io/google-agent-…

Google turns the infrastructure behind Gemini’s agents into a cloud service

Google’s Agent Compute Engine, which already powers Gemini, gives every AI agent its own hardware-isolated microVM. A preview arrives later this year.

thenewstack.io
421
PHP2Shell’s chaotic disclosure sparked a row. Behind the drama, maintainers warn that AI is piling new security work onto already stretched teams. thenewstack.io/php2shell-ai-…

PHP’s security scramble is just a preview: AI is digging up flaws faster than maintainers can fix...

PHP2Shell’s chaotic disclosure sparked a row. Behind the drama, maintainers warn that AI is piling new security work onto already stretched teams.

thenewstack.io
3
1
472
Both companies want to make their AI chips more accessible to PyTorch developers. It’s still hard, though. thenewstack.io/pytorch-tpu-t…

Google and AWS open new paths for PyTorch beyond Nvidia

Both companies want to make their AI chips more accessible to PyTorch developers. It’s still hard, though.

thenewstack.io
384
Letta’s new FrontierMemory benchmark tests whether AI agents learn from experience, and the harness matters as much as the model behind it. thenewstack.io/letta-frontie…

Five AI agent harnesses took the same memory test. The results weren't even close.

Letta’s new FrontierMemory benchmark tests whether AI agents learn from experience, and the harness matters as much as the model behind it.

thenewstack.io
350
OpenAI reports dramatic improvements in legacy software modernization, but it's keeping the tests behind its biggest claim private. thenewstack.io/penai-cobol-m…

OpenAI’s COBOL migration score has made a striking leap

OpenAI reports dramatic improvements in legacy software modernization, but it's keeping the tests behind its biggest claim private.

thenewstack.io
359
OpenAI is donating Open Responses, its spec for a common LLM API, to the Agentic AI Foundation. Here’s who supports it and who’s still missing. thenewstack.io/openai-donate…

OpenAI donates Open Responses, its spec for a common LLM API, to the Agentic AI Foundation

OpenAI is donating Open Responses, its spec for a common LLM API, to the Agentic AI Foundation. Here’s who supports it and who’s still missing.

thenewstack.io
377
Ramp built its own AI agent harness, Laminar, instead of buying one. LangChain’s Harrison Chase sees a bigger shift in how agents get built. thenewstack.io/ramp-laminar-…

Ramp built an AI agent harness in six weeks — and LangChain's CEO says it's part of a bigger shift

Ramp built its own AI agent harness, Laminar, instead of buying one. LangChain’s Harrison Chase sees a bigger shift in how agents get built.

thenewstack.io
1
1
401
Two critical PHP flaws sent some sites offline. The fight over Chainguard’s early patch release may outlast the bugs themselves. thenewstack.io/php2shell-cha…

AI found a critical PHP flaw. The bigger failure began in 2024.

Two critical PHP flaws sent some sites offline. The fight over Chainguard’s early patch release may outlast the bugs themselves.

thenewstack.io
434
When Postgres starts showing signs of strain, migration isn't the only option. Catch the on-demand session for a closer look at scaling live data workloads and understanding what actually needs to change. Watch now: bit.ly/4cs3Lt8
325
Haiku 5.5 missed just four of 1,600 fields. Two came from a date it shouldn’t have trusted thenewstack.io/deepseek-flas…

DeepSeek V4.1-Flash-1017 vs. Claude Haiku 5.5: More accurate at a fraction of the price

Haiku 5.5 missed just four of 1,600 fields. Two came from a date it shouldn’t have trusted

thenewstack.io
470
Google’s Agent Compute Engine, which already powers Gemini, gives every AI agent its own hardware-isolated microVM. A preview arrives later this year. thenewstack.io/google-agent-…

Google turns the infrastructure behind Gemini’s agents into a cloud service

Google’s Agent Compute Engine, which already powers Gemini, gives every AI agent its own hardware-isolated microVM. A preview arrives later this year.

thenewstack.io
2
1
478
Running AI agents on Kubernetes is turning routine. But agents expect sub-200 ms replies, and a new pod is only guaranteed to start in about a second. Thanks to HPE thenewstack.io/kagent-kubern…

AI agents turn one request into about 20. Kubernetes is evolving to handle it.

Running AI agents on Kubernetes is turning routine. But agents expect sub-200 ms replies, and a new pod is only guaranteed to start in about a second.

thenewstack.io
379
Load more