News

The New Stack
thenewstack. io > coding-agent-endurance-gap

Xiaomi's Mi Mo Code claims it beats Claude Code past 200 steps

2+ hour, 23+ min ago  (667+ words) A coding agent that scaffolds a working app over lunch will routinely stall around 30 steps into a production refactor. It locks onto a hypothesis early and keeps patching a wrong assumption, so small errors compound until the run comes apart....

Symbols: btc-usd,eth-usd
The New Stack
thenewstack. io > audit-trails-revenue-asset

What your logs can't tell you when an AI agent acts alone

3+ hour, 23+ min ago  (671+ words) AI agents are acting autonomously, making basic logs obsolete. Discover why comprehensive audit trails are now a critical revenue asset....

Symbols: nyse:crm,d05.S0,u11.S0,z74.S0,cin.si,584.S0
The New Stack
thenewstack. io > tensors-beyond-vector-search

Why AI retrieval and ranking need more than vector search

1+ day, 1+ hour ago  (23+ words) A Giga Om Cx O brief explains why production AI retrieval needs more than vector search and explores how tensors unify ranking and ML signals....

Symbols: nasdaq:googl
The New Stack
thenewstack. io > jetbrains-course-creators-program

Can Jet Brains close the IDE skills gap before AI widens it further?

1+ day, 2+ hour ago  (250+ words) Jet Brains' new Course Creators Program lets educators embed hands-on coding into its IDEs, arguing AI makes foundational developer skills matter more....

Symbols: btc-usd,eth-usd,xrp-usd
The New Stack
thenewstack. io > agent-loops-cloud-native-verification

Loops are replacing prompts. Verification is about to be your biggest problem.

1+ day, 3+ hour ago  (1101+ words) As AI coding shifts from prompts to loops, verification becomes the ultimate challenge for cloud-native engineering teams....

The New Stack
thenewstack. io > fable-5-opus-comparison

Fable 5 vs Opus 4. 8: The real stakes, not the spec sheet

1+ day, 4+ hour ago  (349+ words) Anthropic's new Fable 5 promised a step change over Opus 4. 8. We ran identical coding and reasoning tests. They converged " but the bills didn't....

Symbols: anth.pvt,btc-usd
The New Stack
thenewstack. io > claude-fable-cost-model-triage

Claude Fable cost $9 in one coding test. GPT-5. 5 cost $1. 50. Model triage is the new AI skill.

1+ day, 8+ hour ago  (31+ words) Anthropic's two-week Fable window, Citadel's tokenomics warning, and Open AI's looming price cuts all point the same way: the hard part isn't using the best model. It's knowing when not to....

Symbols: nasdaq:msft
The New Stack
thenewstack. io > us-gov-orders-anthropic-to-pull-fable-5-and-mythos-5-three-days-after-launch

Federal government orders Anthropic to pull Fable 5 and Mythos 5, three days after launch

1+ day, 16+ hour ago  (573+ words) Matt Burns is Director of Editorial at Insight Media Group, where he oversees The New Stack, Roadmap. sh, and Towards Data Science " three platforms that collectively help millions of developers figure out what to learn next. Previously, he spent 16 years…...

Symbols: btc-usd
The New Stack
thenewstack. io > outsystems-agent-orchestration-neutrality

Who gets to be Switzerland in the enterprise agent wars?

1+ day, 23+ hour ago  (554+ words) Every vendor is selling AI agent orchestration. Out Systems CEO Woodson Martin argues neutrality " not owning the data " is the real edge....

Symbols: d05.S0,u11.S0,z74.S0,579.S0,5wh.si,re4.S0
The New Stack
thenewstack. io > stack-overflow-for-agents

Coding agents have questions, too " so Stack Overflow built them a home

2+ day, 2+ hour ago  (687+ words) Stack Overflow has been the internet's go-to troubleshooting ground for software developers for more than 15 years " the place where a 4. a. m production crisis would meet a community that has seen it all before. But the audience it was built for…...

Symbols: btc-usd