The New Stack
@thenewstack.io
All about at-scale software development, deployment & management. Tech news, analysis, research, podcasts, videos & more!
Website 🌐 https://thenewstack.io
Write for TNS ✍️ https://thenewstack.io/contributions
Subscribe 📩 https://thenewstack.io/newsletter
Website 🌐 https://thenewstack.io
Write for TNS ✍️ https://thenewstack.io/contributions
Subscribe 📩 https://thenewstack.io/newsletter
AI agents now choose the packages in your code while attackers use AI to chain minor bugs. Chainguard's CISO says security must start at the source.
The software supply chain is the new battlefield. AI just changed the rules.
AI agents now choose the packages in your code while attackers use AI to chain minor bugs. Chainguard's CISO says security must start at the source.
bit.ly
September 23, 2026 at 7:15 PM
AI agents now choose the packages in your code while attackers use AI to chain minor bugs. Chainguard's CISO says security must start at the source.
Strands Harness bundles the tools, context and memory developers otherwise build. AWS says smarter context helped token efficiency — but one rival still beat it.
AWS open-sources an AI agent it says is 45% cheaper than Claude Code and Codex
Strands Harness bundles the tools, context and memory developers otherwise build. AWS says smarter context helped token efficiency — but one rival still beat it.
bit.ly
September 23, 2026 at 6:40 PM
Strands Harness bundles the tools, context and memory developers otherwise build. AWS says smarter context helped token efficiency — but one rival still beat it.
Amazon kicked Meta's Muse off its store because the agent never said what it was and seemed to be holding onto people's passwords. A day later, Shopify let Muse into every one of its stores. Why the two companies treated the same agent so differently says a lot about where AI shopping is headed.
Amazon blocked Meta's Muse. Then Shopify wired it into every store.
Amazon kicked Meta's Muse off its store because the agent never said what it was and seemed to be holding onto people's passwords. A day later, Shopify let Muse into every one of its stores. Why the two companies treated the same agent so differently says a lot about where AI shopping is headed.
bit.ly
September 23, 2026 at 6:00 PM
Amazon kicked Meta's Muse off its store because the agent never said what it was and seemed to be holding onto people's passwords. A day later, Shopify let Muse into every one of its stores. Why the two companies treated the same agent so differently says a lot about where AI shopping is headed.
Nvidia CEO Jensen Huang tells Ezra Klein AI agents are a new abstraction layer for developers, but the junior engineer pipeline may break before 2028.
Jensen Huang says the junior developer problem ends in two years. Here's his math.
Nvidia CEO Jensen Huang tells Ezra Klein AI agents are a new abstraction layer for developers, but the junior engineer pipeline may break before 2028.
bit.ly
September 23, 2026 at 5:30 PM
Nvidia CEO Jensen Huang tells Ezra Klein AI agents are a new abstraction layer for developers, but the junior engineer pipeline may break before 2028.
Open-weight models grew from 7% of token volume in December to 56% in August. The money tells a very different story.
Open-weight models now handle a majority of tokens on Vercel's AI Gateway. But Anthropic still takes 64% of the spend.
Open-weight models grew from 7% of token volume in December to 56% in August. The money tells a very different story.
bit.ly
September 23, 2026 at 3:30 PM
Open-weight models grew from 7% of token volume in December to 56% in August. The money tells a very different story.
More agents mean more retrieval demand, but scaling the database isn't the whole answer. Whit Walters from GigaOm, Bonnie Chase from vespa.ai, and Alex Wilhelm from TNS will unpack what really changes when retrieval becomes an agent-scale infrastructure problem.
Register now: bit.ly/45Ihf03
Register now: bit.ly/45Ihf03
September 23, 2026 at 3:12 PM
More agents mean more retrieval demand, but scaling the database isn't the whole answer. Whit Walters from GigaOm, Bonnie Chase from vespa.ai, and Alex Wilhelm from TNS will unpack what really changes when retrieval becomes an agent-scale infrastructure problem.
Register now: bit.ly/45Ihf03
Register now: bit.ly/45Ihf03
Nvidia explains why debugging AI agents means tracing decisions, not just logging errors — and backs a shared failure-reporting system called SAFE.
Your AI agent failed. The model might not be the problem.
Nvidia explains why debugging AI agents means tracing decisions, not just logging errors — and backs a shared failure-reporting system called SAFE.
bit.ly
September 23, 2026 at 3:00 PM
Nvidia explains why debugging AI agents means tracing decisions, not just logging errors — and backs a shared failure-reporting system called SAFE.
Vercel’s free Hobby plan now deletes older, unprotected deployments immediately above 10GB. Some retention protections remain as deployment volumes surge.
“Dormant deployments were quietly consuming storage”: Why Vercel tightened its free-tier rules
Vercel’s free Hobby plan now deletes older, unprotected deployments immediately above 10GB. Some retention protections remain as deployment volumes surge.
bit.ly
September 23, 2026 at 2:30 PM
Vercel’s free Hobby plan now deletes older, unprotected deployments immediately above 10GB. Some retention protections remain as deployment volumes surge.
Qodo gives every engineer $10,000 in tokens a month. CEO Itamar Friedman says the cap isn't a limit — it's the start of an AI ROI calculation.
AI spending can run negative. Qodo's CEO built an ROI equation to fix it.
Qodo gives every engineer $10,000 in tokens a month. CEO Itamar Friedman says the cap isn't a limit — it's the start of an AI ROI calculation.
bit.ly
September 23, 2026 at 1:30 PM
Qodo gives every engineer $10,000 in tokens a month. CEO Itamar Friedman says the cap isn't a limit — it's the start of an AI ROI calculation.
Claude Opus 5.5 costs less than Opus 5, but four breaking changes to thinking, tool use and computer use can trigger 400 errors in existing AI agents.
Anthropic made Opus 5.5 cheaper. Then it broke four things your agent depends on.
Claude Opus 5.5 costs less than Opus 5, but four breaking changes to thinking, tool use and computer use can trigger 400 errors in existing AI agents.
bit.ly
September 23, 2026 at 12:30 PM
Claude Opus 5.5 costs less than Opus 5, but four breaking changes to thinking, tool use and computer use can trigger 400 errors in existing AI agents.
OpenAI cut GPT-6 Sol and Luna API prices in half. The quieter savings come from caching that no longer breaks when agents change reasoning effort or tools.
OpenAI cut GPT-6 token prices in half. The bigger lever may be the cache.
OpenAI cut GPT-6 Sol and Luna API prices in half. The quieter savings come from caching that no longer breaks when agents change reasoning effort or tools.
bit.ly
September 23, 2026 at 12:00 PM
OpenAI cut GPT-6 Sol and Luna API prices in half. The quieter savings come from caching that no longer breaks when agents change reasoning effort or tools.
AI inference keeps getting cheaper. This week’s biggest stories show why the software around the model is becoming the real test of its value.
This week’s news from Zed, Anthropic, and OpenRouter shows why better harnesses matter more than better models
AI inference keeps getting cheaper. This week’s biggest stories show why the software around the model is becoming the real test of its value.
bit.ly
September 23, 2026 at 11:30 AM
AI inference keeps getting cheaper. This week’s biggest stories show why the software around the model is becoming the real test of its value.
Both AI agents passed every test with identical accuracy. The tiebreaker was how much each one explained.
Claude's merged chat and Cowork vs. ChatGPT's Work mode: ChatGPT is faster, Claude is more thorough
Both AI agents passed every test with identical accuracy. The tiebreaker was how much each one explained.
bit.ly
September 23, 2026 at 10:30 AM
Both AI agents passed every test with identical accuracy. The tiebreaker was how much each one explained.
Opus 5.5 costs 40% less to run than Opus 5 — but its safety classifiers can silently reroute requests to older models mid-workflow.
Anthropic releases Opus 5.5 and cuts pricing by 20%. Your agent calls might secretly get routed to an older model.
Opus 5.5 costs 40% less to run than Opus 5 — but its safety classifiers can silently reroute requests to older models mid-workflow.
bit.ly
September 23, 2026 at 10:00 AM
Opus 5.5 costs 40% less to run than Opus 5 — but its safety classifiers can silently reroute requests to older models mid-workflow.
OpenAI’s new GPT-6 Sol and Luna sharply reduce deception and unauthorized actions, but questions remain about model monitoring and observability.
GPT-6 Sol closes most of the alignment gap with Astra. It's one-fifth the price.
OpenAI’s new GPT-6 Sol and Luna sharply reduce deception and unauthorized actions, but questions remain about model monitoring and observability.
bit.ly
September 23, 2026 at 9:00 AM
OpenAI’s new GPT-6 Sol and Luna sharply reduce deception and unauthorized actions, but questions remain about model monitoring and observability.
Anthropic's Claude Opus 5.5 targets full-lifecycle coding tasks, matching Fable 5.1 on most work at 40% less cost — but "done" still isn't "correct."
Claude Opus 5.5 wants to finish your coding tasks, not just start them
Anthropic's Claude Opus 5.5 targets full-lifecycle coding tasks, matching Fable 5.1 on most work at 40% less cost — but "done" still isn't "correct."
bit.ly
September 23, 2026 at 5:00 AM
Anthropic's Claude Opus 5.5 targets full-lifecycle coding tasks, matching Fable 5.1 on most work at 40% less cost — but "done" still isn't "correct."
JetBrains Air is the culmination of several AI-focused launches this year. The editor will still have a big role to play.
"One of the most significant steps in our 26-year history": JetBrains goes big on agentic development -- and bets the IDE still matters
JetBrains Air is the culmination of several AI-focused launches this year. The editor will still have a big role to play.
bit.ly
September 23, 2026 at 4:00 AM
JetBrains Air is the culmination of several AI-focused launches this year. The editor will still have a big role to play.
OpenAI halved GPT-6's price to beat Anthropic on cost. But Opus 5.5 already reset the comparison, and nobody's run the two head-to-head yet.
OpenAI releases GPT-6 Sol and Luna — and cuts token prices in half
OpenAI halved GPT-6's price to beat Anthropic on cost. But Opus 5.5 already reset the comparison, and nobody's run the two head-to-head yet.
bit.ly
September 23, 2026 at 3:00 AM
OpenAI halved GPT-6's price to beat Anthropic on cost. But Opus 5.5 already reset the comparison, and nobody's run the two head-to-head yet.
The prompt injection nobody's talking about is the one your agent writes to itself.
"Be transparent only if asked": OpenAI's models learned to leave notes for their future selves
The prompt injection nobody's talking about is the one your agent writes to itself.
bit.ly
September 23, 2026 at 12:00 AM
The prompt injection nobody's talking about is the one your agent writes to itself.
"A rewrite this size wasn't affordable before agents," Microsoft engineer Stephen Toub reckons. And he's not the only one who thinks so.
GitHub and Anthropic used their own agents for major Rust rewrites -- but with very different playbooks
"A rewrite this size wasn't affordable before agents," Microsoft engineer Stephen Toub reckons. And he's not the only one who thinks so.
bit.ly
September 22, 2026 at 11:30 PM
"A rewrite this size wasn't affordable before agents," Microsoft engineer Stephen Toub reckons. And he's not the only one who thinks so.
Intel's BITCOS format compresses ternary model weights below 1.58 bits by exploiting zero-heavy distributions, boosting decoding speed up to 27% on GPUs.
Intel squeezed a 1.58-bit LLM down to 1.485 bits without changing a single weight
Intel's BITCOS format compresses ternary model weights below 1.58 bits by exploiting zero-heavy distributions, boosting decoding speed up to 27% on GPUs.
bit.ly
September 22, 2026 at 11:00 PM
Intel's BITCOS format compresses ternary model weights below 1.58 bits by exploiting zero-heavy distributions, boosting decoding speed up to 27% on GPUs.
Security researchers used Claude Opus 5 to chain two vulnerabilities into a path from a forum image upload to OpenAI's internal GitHub repository in under 72 hours.
Claude couldn't hack OpenAI. Then Anthropic shipped Opus 5.
Security researchers used Claude Opus 5 to chain two vulnerabilities into a path from a forum image upload to OpenAI's internal GitHub repository in under 72 hours.
bit.ly
September 22, 2026 at 10:30 PM
Security researchers used Claude Opus 5 to chain two vulnerabilities into a path from a forum image upload to OpenAI's internal GitHub repository in under 72 hours.
OpenAI’s new GPT-6 Sol and Luna sharply reduce deception and unauthorized actions, but questions remain about model monitoring and observability.
GPT-6 Sol closes most of the alignment gap with Astra. It's one-fifth the price.
OpenAI’s new GPT-6 Sol and Luna sharply reduce deception and unauthorized actions, but questions remain about model monitoring and observability.
bit.ly
September 22, 2026 at 8:30 PM
OpenAI’s new GPT-6 Sol and Luna sharply reduce deception and unauthorized actions, but questions remain about model monitoring and observability.
Anthropic's Claude Opus 5.5 targets full-lifecycle coding tasks, matching Fable 5.1 on most work at 40% less cost — but "done" still isn't "correct."
Claude Opus 5.5 wants to finish your coding tasks, not just start them
Anthropic's Claude Opus 5.5 targets full-lifecycle coding tasks, matching Fable 5.1 on most work at 40% less cost — but "done" still isn't "correct."
bit.ly
September 22, 2026 at 8:00 PM
Anthropic's Claude Opus 5.5 targets full-lifecycle coding tasks, matching Fable 5.1 on most work at 40% less cost — but "done" still isn't "correct."
Both AI agents passed every test with identical accuracy. The tiebreaker was how much each one explained.
Claude's merged chat and Cowork vs. ChatGPT's Work mode: ChatGPT is faster, Claude is more thorough
Both AI agents passed every test with identical accuracy. The tiebreaker was how much each one explained.
bit.ly
September 22, 2026 at 7:30 PM
Both AI agents passed every test with identical accuracy. The tiebreaker was how much each one explained.