Install
The New Stack is a media platform for the people who build and manage software the world relies on. We provide context and explanation of at-scale technologies to advance knowledge and create conversations through our coverage of modern architectures, components of the software development life cycle, and operations to
- 152articles · 30d
- 6+ hour agolatest article
- Aug 14, 2026earliest in window
- 96%with images
- 88avg words
- Science & Technology 140
- Software Dev. 101
- Computers & Electronics 95
- News 38
- Software 21
- Science & Nature 13
- Economy, Business & Finance 9
- Finance & Business 9
Please confirm you are human
This browser or connection looks automated. Press and continuously hold the control for 3 seconds to enable Google-hosted web results and, when separately allowed, AI-assisted answers.
A successful check enables 100 search requests. Interactive access does not authorize scraping, systematic collection, or reuse of search output.
News
“Machine translation is still broken for most of the world's languages”: Cohere builds non-reasoning for a reason
7+ hour, 20+ min ago (203+ words) Cohere's North Small Translate beats DeepL and Google Translate on WMT26 across 50 languages — but commercial use requires a Model Vault license....
Claude did best on a new benchmark for agents that build agents. It still passed fewer than a quarter of the tests.
4+ day, 1+ hour ago (492+ words) Sierra has open-sourced Hyper-𝜏-bench, a follow-up to its 2024 τ-bench that tests how well AI agents can build other agents....
K2 Horizon just shipped as six new fully open models — developers aren't fully convinced
4+ day, 9+ hour ago (632+ words) Based in the Emirati capital, Abu Dhabi, the Institute of Foundation Models (IFM) introduced K2 Horizon last week. This group of six AI foundation models, ranging from 0.9 billion to 375 billion parameters, is claimed to be the “largest fully open-source fleet of…...
Polars 2.0 pre-release comes with a 5x speed boost — but it could change row order
1+ week, 8+ hour ago (81+ words) The streaming engine becomes the default for LazyFrame queries, improving memory and performance. There’s one hiccup, though....
Cut GPU inference cold start from 8 minutes to less than a minute
1+ week, 3+ day ago (635+ words) Cut GPU inference cold start times from 8 minutes to under 30 seconds with simple configuration and platform fixes....
Want to scale AI agents without breaking anything? Retrieval engineering is the answer.
1+ week, 3+ day ago (23+ words) On September 24, GigaOm’s Whit Walters and Vespa.ai’s Bonnie Chase will explain what breaks when hundreds of AI agents hit retrieval at once....
GLM-5.3-Flash vs. GLM-5.3: Time and money, not the spec sheet
1+ week, 5+ day ago (801+ words) I always wonder about the end goal when companies launch products so close together and undercut each other by claiming the new one is “so much better.” GLM-5.3 and GLM-5.3-Flash are a great example of this. Z.AI launched…...
Meta just beat OpenAI and Google at real-time transcription
1+ week, 5+ day ago (529+ words) Meta’s Superintelligence Labs on Tuesday launched Muse Voice Transcribe, a new real-time speech recognition model that, at least on some benchmarks, outperforms virtually every other comparable model when it comes to working with speech in real time. Meta’s lab describes…...
Cut coding agent token use with better tool output
1+ week, 6+ day ago (695+ words) Before an AI coding agent writes a single line of code, it has already spent tokens. For example, on source files, ticket descriptions, build logs, quality findings, and dependency alerts. Most of what you pay for isn’t the pull request;…...
DeepSeek's first vision model vs. Gemini 3.7 Flash: It comes down to spend vs. speed
1+ week, 6+ day ago (334+ words) DeepSeek matched Gemini on nine vision questions for a third of the price. The tradeoff: slower, less consistent responses that matter at scale....