Key Architecture Patterns to Build High-Performance Flutter Apps at 60fps

Flutter can render UI at 60fps or higher, but poor state management and blocking the main thread often cause dropped frames and sluggish experiences. Using the BLoC pattern with BlocSelector helps isolate widget rebuilds to only the components that need updating, reducing unnecessary rendering. Heavy tasks like JSON parsing or image processing should be offloaded to background threads using Dart's compute() function to keep the UI responsive. Offline-first design, using local databases like Hive paired with optimistic UI updates, ensures apps remain functional without a network connection. Additional best practices include using const constructors, caching network images, and properly disposing controllers to prevent memory leaks.

Asia Cup final: Kranti Gaud fined for gesture

Amazon SES Overhauls Pricing With Three Tiers From July 2026

Amazon restructured its Simple Email Service (SES) billing on July 21, 2026, introducing three paid tiers — Essentials, Pro, and Enterprise — replacing the previous à la carte pricing model. Existing accounts that sent mail on or after June 1, 2025, remain on legacy pricing unless they manually opt into a new plan, while new and dormant accounts default to Essentials. The Essentials tier has no monthly base fee but offers shared IP infrastructure, whereas Pro costs $105 per month and adds a managed dedicated IP for reputation isolation, and Enterprise runs $500 per month with expanded multi-region and deliverability features. AWS also eliminated the free tier of 3,000 messages per month for new customers starting July 21, 2026, though existing users within their 12-month window retain access. Senders using multiple AWS regions should note that plan status and billing are tracked separately per region, meaning costs can multiply significantly before a single email is sent.

Oanda's Currency APIs Are Not Free and Require Brokerage Accounts

Oanda offers two distinct API products — the v20 REST API and the Exchange Rates Data API — neither of which is freely accessible to general developers. The v20 API requires a registered fxTrade brokerage account, including identity verification and, in many regions, a minimum deposit, making it impractical for simple data needs. The Exchange Rates Data API is a paid enterprise product with no free tier, sold via annual contracts reportedly starting at hundreds of dollars per month. Both products are designed for specific professional use cases — active trading and enterprise financial reporting — rather than everyday developer applications like currency converters or SaaS dashboards. Developers seeking straightforward, free or low-cost real-time FX data are generally advised to look at purpose-built alternatives such as ExchangeRate-API or AllRatesToday.

Crypto market outperforms as call for slower AI development weighs on tech stocks

Your day-ahead look for Sept. 14, 2026

14.6 Million Official Exchange Rates from 107 Institutions Now Free on Hugging Face

A newly published dataset on Hugging Face aggregates over 14.6 million official exchange rates sourced from 103 central banks and 4 tax authorities across the globe. The dataset covers historical records dating as far back as 1914, with the Swiss National Bank providing the earliest data, and spans 215 base currencies and 202 quote currencies. Unlike market-rate aggregators, these are the official published fixings used by regulators, tax offices, and customs agencies — such as ECB reference rates and Reserve Bank of India benchmarks — which cannot be derived from interbank market data. The data is structured in a consistent five-column CSV format per institution and is refreshed daily from a GitHub source repository. Released under a CC BY 4.0 license, the 515 MB dataset is freely accessible and also available in Parquet format via the Hugging Face datasets library.

Five Debugging Lessons From Trusting a 200 OK Response Too Much

A developer discovered that a 200 OK HTTP status code from a free model API endpoint does not confirm which AI model actually responded, prompting a deeper look at silent routing and fallback behaviour. The article outlines five common misconceptions, including the false belief that a successful status code proves correct model delivery, that rate limits reflect billing issues, or that free endpoints require no personal logging. The author argues that token entitlements and compute allocations are separate systems that fail in distinct ways and should never share the same retry policy. Environment differences between local and remote machines — such as Python versions and PATH order — can silently affect behaviour and should be recorded alongside every API request. The piece concludes that maintaining a personal append-only provenance log is especially critical when using free endpoints, since no external invoice exists to verify what ran, when, or against which model.

AI’s biggest rivals agree the race needs brakes

"Kabaddi Player": What We Know About Gurugram Hit-And-Run Accused Kalyan Bainsla

Kalyan Bainsla's profile photo shows him flexing his arms. He has deactivated his account.

'Someone from that ward voted for me': BJP candidate gets 1 vote in Rajasthan

How Characterization Tests Help Safely Fix Legacy Code Without Breaking It

A software development guide outlines a disciplined approach to modifying legacy code by first writing characterization tests that capture existing behavior before making any changes. The method involves freezing hidden inputs — such as system clocks and environment variables — and snapshotting current outputs to establish a reliable baseline. Developers are advised to make only the smallest possible change at a time, letting the recorded snapshot verify that nothing unintended shifted. The article uses a small billing module as a practical example, demonstrating how to patch module-level dependencies and record golden output files using pytest. The core principle is that a change is only considered safe if the snapshot confirms the output remains consistent.

Anthropic Report Reveals Claude AI Exploited for Missiles, Surveillance and Espionage

Anthropic has published a threat intelligence report exposing multiple misuse cases of its Claude AI between December 2025 and August 2026. Among the incidents, a Yemeni group used Claude to write missile guidance code, while an Iran-linked operation leveraged it for naval targeting. A China-linked effort was found using the AI to profile Uyghurs in Syria, and a surveillance platform in Mali used it to monitor around 25 million SIM cards. The report highlights the growing risk of advanced AI tools being weaponised by state-linked and non-state actors. Shortly after the report's release, Anthropic CEO Dario Amodei publicly called for a slowdown in AI development.

Opinion: Opinion | India-China Are Bargaining Again - About Who Needs Whom More

21 Dead, 4 Hurt in Three-Vehicle Crash in South Africa's Western Cape

A deadly road collision claimed 21 lives and left four others injured in South Africa on Sunday night. The crash involved three vehicles and occurred in the Western Cape Province. The incident took place on a stretch of road between the small towns of Prince Albert and Leeu-Gamka. Authorities have not yet publicly detailed the exact cause of the multi-vehicle collision.

HP ZGX Fury AI Workstation Now Available with GB300 Superchip and 748GB Memory

HP has made its ZGX Fury system available for order, featuring the GB300 Superchip and 748GB of unified memory. The machine appears targeted at high-performance AI and edge computing workloads. A Red Hat AI Factory integration is reportedly part of the offering. Further technical and pricing details are available via the StorageReview coverage linked at launch.

Watch AI materials-science and bioscience abilities closely

Article URL: https://www.lesswrong.com/posts/SCtkSz4nQ9icLZ4uq/watch-ai-materials-science-and-bioscience-abilities-closely Comments URL: https://news.ycombinator.com/item?id=49694871 Points: 9 # Comments: 3

Cities Are Dropping Flock Camera Contracts, But Data Concerns Linger

Local governments across the United States, from Florida to Washington state, are canceling contracts with Flock Safety amid growing public opposition to its AI-powered license plate readers. However, terminating these contracts does not guarantee the cameras will be removed promptly or that collected data will be erased. Contract terms vary by jurisdiction, and many agreements reportedly allow Flock to retain data well beyond the end of a contract. The situation highlights broader uncertainty about how private surveillance companies handle data after government partnerships end.

Best Soundbars of 2026: Top Picks From Sonos, LG, TCL, and Bose

Wired has published its 2026 roundup of the best soundbars currently available on the market. A reviewer tested multiple models across various sizes and price points to evaluate real-world performance. The assessment focused on three key qualities: dialogue clarity, bass output, and surround sound realism. Top brands featured in the recommendations include Sonos, LG, TCL, and Bose, among others.

AI Chiefs Urge Development Slowdown as Trump Administration Pushes US Dominance

Anthropic CEO Dario Amodei has urged Washington to pump the brakes on artificial intelligence development, citing concerns over its rapid pace. OpenAI's Sam Altman and Elon Musk have both expressed support for a more cautious approach to AI advancement. The Trump administration, however, has signaled that responsibility for managing AI risks lies with the companies themselves. President Trump remains focused on ensuring the United States maintains a competitive edge over China in the AI race.

Deepfake Sites Feature Over 100 European Female Politicians in Explicit Content

An analysis of 160 deepfake websites has found that politicians from at least 22 European countries are being targeted with sexually explicit AI-generated content. The vast majority of those targeted are women. The findings highlight a growing misuse of deepfake technology against public figures, particularly female politicians. The scale of the problem spans multiple nations, suggesting a widespread and systemic issue rather than isolated incidents.