Accelerating GPT-5.6 Sol Ultrafast
• Cerebras powers GPT-5.6 Sol Ultrafast mode, delivering up to 750 output tokens per second.Read more | Comments | @hackernws
• Ultrafast mode runs 11x faster than Fable 5 and 5x faster than Opus 4.8 Fast mode.
• GPT-5.6 Sol Ultrafast answered 2,500 PhD-level questions in 11 hours, beating Claude Fable 5's 78 hours.
• Ultrafast achieved a 5.6x speedup on economically valuable tasks without any quality loss.
• Ultrafast uses Cerebras’ Wafer-Scale Engine with 44 GB SRAM to eliminate data movement bottlenecks.#GPT56 #AIAcceleration #Cerebras
www.cerebras.ai
Accelerating GPT-5.6 Sol Ultrafast with OpenAI
Cerebras powers OpenAI’s GPT-5.6 Sol Ultrafast in the OpenAI API, delivering frontier intelligence at real-time speeds for critical AI work.
Mistral OCR 4.1
• OCR 4.1 is Mistral AI's latest OCR service with paragraph-level bounding box extraction.Read more | Comments | @hackernws
• The model provides structural block labels and block-level confidence scores.
• Pricing is €3.5 per 1000 pages and €4.38 per 1000 annotated pages.
• Available via API with features like batching and structured annotations.
• Released in public preview on July 16, 2026.#OCR #MistralAI #DocumentAI
docs.mistral.ai
OCR 4.1 - Mistral AI
Our latest OCR service powering our Document AI stack, with native paragraph-level bounding box extraction, structural block labels, and block-level confidence scores.
Choosing an AI model: one prompt, 11 models, different results
• Netlify partnered with OpenRouter to offer 11 AI models via AI Gateway and Agent Runners.Read more | Comments | @hackernws
• AXIS tool evaluated models on building and iterating three types of web projects.
• Claude Opus 5 used most credits (average 519), delivering detailed, complex designs.
• DeepSeek V4 Flash consumed least credits (average 2.4) with mixed results.
• Open and closed models differ in cost, design quality, and handling of platform features.#AIModels #Netlify #OpenRouter
Netlify
More models, more choice: Comparing 11 different AI models
Netlify now runs any OpenRouter model, including Kimi K3, GLM 5.2 and DeepSeek V4. We tested 11 of them on the same build prompt to see how they differ.
Ordinary abundance
• Electric lighting fully illuminated Wabash, Indiana in 1880, a major advancement over candles.Read more | Comments | @hackernws
• Edward Jenner's smallpox vaccine greatly reduced deaths, cited by Thomas Jefferson in 1806.
• Anaesthesia introduction in 1846 revolutionized surgery by eliminating patient pain.
• Indoor plumbing became common in rural America only in the early 20th century.
• The safety bicycle, praised by Susan B. Anthony in 1896, advanced women's mobility and independence.#InnovationHistory #EverydayTech #HumanProgress
Ordinary Abundance
A walk through a modern apartment, through the eyes of the people for whom everything in it was new.
Choose Boring Technology (2015)
• Choosing boring, well-understood technology reduces unknown failures and operational overhead.Read more | Comments | @hackernws
• Companies have limited 'innovation tokens' to spend on new technology choices.
• Adding new tech requires company-wide discussion and clear migration plans.
• Boring but good tech like MySQL, Python offers stability and scalability over shiny new tools.
• Etsy's experience shows conservative tech choices support scalable, maintainable growth.#TechStrategy #SoftwareEngineering #InnovationManagement
Dan McKinley :: Math, Programming, and Minority Reports
Dan McKinley :: Choose Boring Technology
Antiqua–Fraktur dispute
• The Antiqua–Fraktur dispute was a German typographical debate from the 19th to early 20th century.Read more | Comments | @hackernws
• Fraktur (blackletter) was deemed more 'German' and serious, Antiqua was seen as 'un-German'.
• In 1911, a Reichstag vote narrowly rejected making Antiqua the official typeface.
• Nazi regime initially promoted Fraktur but banned it in 1941 for its 'Jewish letters' stigma.
• Post-WWII, Latin script dominated; Fraktur persists in limited cultural and religious contexts.#Typography #GermanHistory #Fraktur
Wikipedia
Antiqua–Fraktur dispute
typographical dispute in 19th- and early 20th-century Germany
Donkey.bas is 45 Years Old – 131 line of Glory
• DONKEY.BAS is a 1981 IBM PC game demoing color graphics and sound in BASICA.Read more | Comments | @hackernws
• It was co-written by Bill Gates and Neil Konzen.
• The game involves switching lanes to avoid hitting a donkey.
• The page offers a JavaScript version replicating original CGA gameplay.
• Original source code is available on GitHub and donkeybas.com.#DONKEYBAS #IBMPC #RetroGaming
Donkeybas
DONKEY.BAS — IBM PC (1981)
Play DONKEY.BAS in your browser — the classic IBM PC game by Bill Gates and Neil Konzen, for the 45th anniversary of the IBM PC.
Kubernetes on Oxide: How customer needs shaped our integrations
• Oxide supports Kubernetes provisioning via Rancher node driver, Omni infrastructure provider, and Cluster API provider.Read more | Comments | @hackernws
• Oxide cloud controller manager (CCM) synchronizes Kubernetes Node objects with Oxide instances.
• LoadBalancer Kubernetes services use Oxide floating IPs with proxy and VIP mode for traffic.
• Oxide storage CSI plugin is in development; disk hot-plug support required for native integration.
• Future work includes autoscaling, external subnet support, and native load balancing on Oxide.#Kubernetes #OxideComputer #CloudIntegration
Oxide Computer Company
Kubernetes on Oxide: How Customer Needs Shaped Our Integrations | Oxide Computer Company
How customer needs shaped Oxide's Kubernetes integrations.
Understanding is the new bottleneck
• AI agents are writing more code, making human understanding the new bottleneck in software projects.Read more | Comments | @hackernws
• Understanding code helps humans actively participate, not just verify, in iterative AI-driven development loops.
• Three techniques to build understanding: code explainer docs, quizzes for comprehension, and interactive micro-worlds.
• Shared collaborative spaces like Notion help teams develop common mental models for efficient communication.
• Alan Kay's 50-year-old vision supports using interactive simulations and AI to augment human understanding, not automate it away.#AIProgramming #CodeUnderstanding #HumanAICollaboration
Geoffreylitt
Understanding is the new bottleneck
Agents can write code faster than we can absorb it. Here's why it still matters for humans to understand what they build — and some techniques for doing that efficiently: explainer docs, quizzes, micro-worlds, and shared spaces.
Single log line is 49KB+ (ext4) / 110KB+ (btrfs) of systemd-journald disk writes
• systemd-journald version 257.9 on Debian 13 with Linux kernel 6.12.57+deb13-amd64 causes excessive IO.Read more | Comments | @hackernws
• VM running journald writing 2 log lines/second does ~50 IOPS, much higher than syslog.
• File system used is XFS, and journald writes logs directly to hard drive causing slowdown.
• Issue reproduces with constant logging stream from haproxy showing frequent disk write spikes.
• Problem linked to journald inefficiency and inflated log file sizes, also prone to corruption on unclean reboots.#systemd #journald #IOPS
GitHub
Excessive IO caused by systemd-journald · Issue #40262 · systemd/systemd
systemd version the issue has been seen with 257.9 Used distribution Debian 13 Linux kernel version used 6.12.57+deb13-amd64 Component systemd-journald Expected behaviour you didn't see Log wri...
NP-Overrated
• NP-hard problems are theoretically intractable but often practically solvable on most inputs.Read more | Comments | @hackernws
• Dependency resolution and type checking rarely encounter worst-case complexity in real scenarios.
• Advanced optimization tools can find provably optimal solutions for scheduling and traveling salesman problems efficiently.
• Algorithmic improvements outpaced hardware gains, achieving up to 450-billion-fold speedup (1991-2015).
• Even SAT and SMT problems are routinely solved at scale, e.g., Amazon handles a billion SMT queries daily.#NPcomplete #Algorithms #Optimization
Where did the old web go? We followed 657,607 links to find out
• 657,607 links from 0.mk (2009-2014) were tested for loading in August 2026.Read more | Comments | @hackernws
• 76.7% of 655,178 crawlable links no longer returned a loading page.
• 55.0% of unique URLs failed network connection; 23.7% returned HTTP errors, mainly 404.
• Popular giants like YouTube and Wikipedia mostly survived; local Macedonian news sites did not.
• AI now manages 0.mk, enabling its relaunch after 17 years offline.#LinkRot #InternetArchive #WebHistory
US conducted mass spying campaign against leftwing and anti-ICE protesters
• US Department of Homeland Security spied on leftwing groups and anti-ICE protesters in Minnesota and New York.Read more | Comments | @hackernws
• Operation Puppet Master targeted a network opposing immigration enforcement in early 2026.
• Undercover agents attended meetings, infiltrated Signal chats, and obtained financial records of unions and nonprofits.
• 15 Minneapolis protesters charged with conspiracy; major groups like AFL-CIO, DSA not charged.
• Investigations followed Trump’s executive order labeling antifa a domestic terrorist organization.#USSurveillance #ImmigrationProtests #DHSOperation
the Guardian
US conducted ‘mass spying campaign’ against leftwing groups and anti-ICE protesters, records reveal
DHS sent undercover agents to meetings and infiltrated chats of groups opposing Trump’s immigration crackdown
SparrowMap – Cameras that watch government vehicles
• SparrowMap volunteers use spare phones to detect and record government vehicles only.Read more | Comments | @hackernws
• Non-government vehicle data is destroyed on-device; only anonymous data sent to server.
• System runs entirely in-browser with no need for accounts or app installation.
• Footage does not leave device; detection processed locally to prevent data leaks.
• Desktop setups available for always-on monitoring with Windows and Linux support.#SparrowMap #OpenData #GovVehicleTracking
Sparrowmap
Cameras that watch government vehicles, run by volunteers, owned by nobody.
GLM-5.3: Frontier Coding with Emergent Cyber Capabilities
• GLM-5.3 released with significant post-training scaling, same base as GLM-5.2.Read more | Comments | @hackernws
• Improves coding ability by 50% on Z.ai Code Bench and leads open-source coding benchmarks.
• Shows emergent cyber capability, doubling GLM-5.2 on exploitation benchmarks like ExploitBench.
• Discovered 2,436 vulnerabilities in 269 projects, oldest flaw from 1981, with 1,097 critical/high severity.
• Weights to be open-sourced in two weeks after safety evaluation and hardening.#AIcoding #CyberSecurity #OpenSourceAI
How Compaction Works in Pi
• Pi triggers compaction when conversation exceeds LLM context window, preventing errors.Read more | Comments | @hackernws
• Compaction summarizes older conversation parts using an LLM, preserving recent messages.
• Pi's compaction prompt requests structured summaries focusing on goals, progress, and decisions.
• Compaction appends a plain text summary to the session, retaining readable and portable context.
• Compaction breaks prompt caching due to changed context prefix but allows continued use afterwards.#PiLLM #AICompaction #CodingAgents
Earendil
How Compaction Works in Pi | EARENDIL
Why compaction is needed for large language models and how Pi implements it
Hello, me. It's been a while
• Author had not posted on this blog for over 14 years before this personal update.Read more | Comments | @hackernws
• Due to responsibilities, the author now fills silence with podcasts, audiobooks, or social media.
• Author is a slow thinker who values long, uninterrupted self-reflection time.
• Work routine involves meetings and focus, leaving no time for slow, quiet thinking.
• Author rediscovered enjoyment of silence and inner voice by stopping habitual audio playback.#Mindfulness #Silence #SelfReflection
The Mech - programming blog - Python, software development, stock exchange, C#
Hello, me. It’s been a while. - The Mech - programming blog
Wow. It’s been over fourteen years since the last time I wrote a post on this blog. In those fourteen years, the world has changed, and so have I. So this isn't going to be a technical post, it's going to be personal. One of those subtle changes that quietly…
Bluesky Protocol Services
• Bluesky launches Bluesky Protocol Services with a new website and developer docs.Read more | Comments | @hackernws
• Jetstream v2 adds Network Replay for past data catch-up with live WebSocket tail.
• Network Replay is stateless on server; requires API token for archive downloads.
• Jetstream v2 instances live at wss://jetstream.us-west.bsky.network and wss://jetstream.us-east.bsky.network.
• New Jetstream SDKs in TypeScript and Go released, plus updated Bluesky TypeScript SDK based on @atproto/lex.#BlueskyProtocol #JetstreamV2 #APIUpdate
AT Protocol
Introducing Bluesky Protocol Services - AT Protocol
A new home for the infrastructure Bluesky runs on the open network — including Jetstream v2 with network replay, a new Jetstream SDK, and a lex-based TypeScript SDK
Why does Opus 5 feel worse to work with?
• Opus 5 is more capable than Opus 4.7, 4.8, and rivals Fable in benchmarks.Read more | Comments | @hackernws
• Opus 5 requires more 'babysitting' as it makes assumptions without asking for clarification.
• Earlier models pause to ask questions and avoid reinterpreting plans without checking.
• Benchmark-driven model training favors bold assumptions to maximize scores.
• Users prefer agents that ask for clarification due to real-life ambiguity and consequences.#Opus5 #AIbenchmarks #CodingAgents
mun-logadan.github.io
Why does Opus 5 feel worse to work with?
A blog
DeepSeek peak/off-peak pricing update
• DeepSeek-V4-Pro launched with major Agent upgrades and strong production gains.Read more | Comments | @hackernws
• Flexible reasoning effort levels introduced for V4-Pro and V4-Flash models.
• Native OpenAI Responses API support added, optimized for Codex with one-click setup.
• V4-Pro is available on app, web, and API; model names unchanged.
• New API pricing, with 50% lower off-peak rates, effective Aug 16, 2026 at 16:00 UTC.#DeepSeekV4Pro #APIpricing #AIagents
Deepseek
DeepSeek-V4-Pro GA Release | DeepSeek API Docs
We’re launching DeepSeek-V4-Pro today! 🚀
Dear people who work at the airport
• The provided URL returns a 403 Forbidden error, blocking access to the article.Read more | Comments | @hackernws
• Article content is inaccessible due to server restrictions or permissions.
• JavaScript and cookies must be enabled to view the webpage content.#AccessDenied #WebErrors #JavaScriptNeeded
Life after SSRI
Dear people who work at the airport
I would like you to know that we passengers are trying our best.
To you, this place makes sense -- you come here every day. You speak the lingo. You kno...
To you, this place makes sense -- you come here every day. You speak the lingo. You kno...
😐1