News

The New Stack
thenewstack.io > google-frozen-gemini-chip

Google just bet its inference future on a chip built for one model

2+ hour, 44+ min ago  (728+ words) Blueconic sets this cookie as a unique identifier for the BlueConic profile. YouTube sets this cookie to manage feature rollout and experimentation. It helps Google control which new features or interface changes are shown to users as part of testing…...

The New Stack
thenewstack.io > kimi-k3-fable-coding-benchmark

Claude Fable 5 vs. Kimi K3: Same results, one-third the cost, 4x slower

6+ hour, 49+ min ago  (617+ words) Moonshot's open-weight Kimi K3 matched Anthropic's Fable 5 on three coding tasks for a third of the cost — but ran four times slower. Here's the data....

Google News
thenewstack.io > agent-platform-portability-contract

Amazon, Microsoft, and Google are converging on the same enterprise agent architecture

7+ hour, 14+ min ago  (439+ words) What mattered was the contract, not the implementation. An application declared what it needed and stayed agnostic of where it ran. Buildpacks began life at Heroku back in 2011. Pivotal and Heroku started the Cloud Native Buildpacks project in January 2018, and…...

The New Stack
thenewstack.io > fable-5-permanent-subscription-access

Anthropic employees worked "literally around the clock" to keep Fable 5 from disappearing

8+ hour, 20+ min ago  (90+ words) Anthropic finalizes Claude Fable 5 as a permanent Max and Team Premium feature at 50% usage limits after weeks of temporary extensions and infrastructure buildout....

The New Stack
thenewstack.io > self-healing-gpu-nodes

Self-healing GPU nodes in Kubernetes: What we learned building the EKS node monitoring agent

1+ day, 11+ hour ago  (652+ words) Automate self-healing GPU nodes in Kubernetes. Explore 6 critical architecture lessons from building the AWS EKS Node Monitoring Agent....

The New Stack
thenewstack.io > move-code-review-upstream

Move code review before the code

1+ day, 10+ hour ago  (459+ words) Traditional code review is broken by AI. Learn how shifting review upstream to developer intent scales engineering and saves time....

The New Stack
thenewstack.io > spark-4-2-ai-workloads

Spark 4.2 has a feature that could retire your vector database

1+ day, 8+ hour ago  (204+ words) Spark 4.2 adds vector search, governed metrics, streaming upgrades and deeper Python support, positioning the engine as an AI serving layer....

The New Stack
thenewstack.io > serving-environments-agent-speed

Platform engineering's new job: serving environments at agent speed

2+ day, 10+ hour ago  (668+ words) AI coding agents overwhelm environment workflows. Learn how a serving model scales platform engineering for AI-native code velocity....

The New Stack
thenewstack.io > ai-agent-infrastructure-bottleneck

The bottleneck for AI agents isn't the model anymore. It's the context layer.

2+ day, 9+ hour ago  (477+ words) AI agent bottlenecks aren't model problems; they're infrastructure issues. Discover why reliable agents require robust context layers....

The New Stack
thenewstack.io > grok-opus-coding-tokens

I trust Claude for everything. This test made me rethink that.

2+ day, 11+ hour ago  (508+ words) xAI says Grok 4.5 matches Claude Opus 4.8 on coding for a fraction of the tokens. We ran both on three real Rust jobs and tracked every token....