Home Blog Page 5

Meta Releases Muse Spark 1.3 for Agentic Workflows and Coding 

Sato Kamoto
-
0
Meta Releases Muse Spark 1.3 for Agentic Workflows and Coding 

Meta is pushing hard into the agentic era, and their latest model release makes that incredibly obvious. The social media giant just dropped Muse Spark 1.3 into its Muse Code workspace and Meta Model API.

They are not pitching this as a generic chat upgrade. Instead, this system is specifically optimized to build long-running, multi-agent workflows—the exact kind of autonomous setups that developers actually build but constantly struggle to keep on track over long execution threads.

The pricing structure is where things get really fascinating, and frankly, a bit controversial.

Meta is offering two very distinct tiers for accessing the API. The Standard tier runs a reasonable $1.25 per million input tokens and $4.25 per million output tokens.

But if you are on a tight budget, they have introduced a “Contributor” tier priced at a dirt-cheap $0.10 per million input and $0.20 per million output.

The catch? Meta explicitly states that prompts and outputs from this cheaper tier may be used to improve and train their proprietary models.

It is a classic data-for-discount trade-off that will surely give enterprise legal teams a massive headache.

custom-Muse-Spark-1-3-Duck-ai-image-2026-09-04-17-01

Under the hood, the model aims to reduce the “chattiness” of typical AI coding partners.

According to Meta’s launch blog, tests show that Muse Spark 1.3 is significantly more efficient than its predecessor, using roughly 20 percent fewer tool calls and 25 percent fewer total tokens to accomplish the same engineering tasks.

It also features much better instruction-following over long horizons, meaning it won’t drift from your constraints or forget what it was supposed to be doing five prompts ago.

Muse-Spark-1.3-Benchmark-Scorecard
Muse Spark 1.3 Benchmark Scorecard

One of the scariest parts of building autonomous AI agents is the risk of them doing something irreversible—like deleting a production database, executing bad code, or sending a buggy email to a client.

Meta claims they have significantly improved the model’s calibration here. It is designed to be better at identifying “irreversible actions” and pausing to ask for human confirmation before doing something incredibly stupid.

For those ready to test it out, you can pull the model directly on macOS or Linux by executing a quick curl script through the terminal as outlined on the Meta Developer Portal.

If you are already utilizing the Meta Model API, swapping to the new model is as simple as updating a single model string in your configuration files. Let us see if developers are willing to trade their codebase privacy for those bottom-tier token prices.

The White House Arcade Is Vibecoded Garbage—and That Is Being Generous

Lance Cameron
-
0
The White House Arcade Is Vibecoded Garbage—and That Is Being Generous

WASHINGTON, D.C., September 3, 2026 — The White House just launched a federal government-sponsored video game arcade.

Let that sentence sit for a second.

The actual Executive Branch of the United States government is now operating a website—arcade.gov—dedicated to hosting playable browser games. And the results are exactly as chaotic and politically charged as you would imagine.

The announcement dropped on the official @WhiteHouse X account with a casual 7 million views and a link pointing users to five distinct playable titles hosted on whitehouse.gov/arcade/.

These are not polished indie projects. These are rough, obviously rushed, borderline coherent web games with messaging that pulls zero punches.

The Lineup of Official U.S. Government Games

Here is what taxpayers are technically funding as of this week. Five games, each more bewildering than the last:

  • FLAPPY BILL: A shameless Flappy Bird clone where you pilot a bald eagle through the national mall. The metaphor is not subtle.
  • BUILD THE WALL: A Tetris variant where you “protect the border from the coming horde.” The game mechanics do not seem to fully understand how Tetris actually works, but the message is loud.
  • RIO RUN: A snake-style game where you control a border patrol agent capturing migrants one by one. Captured figures trail behind in bright orange prison jumpsuits, forming increasingly grotesque human centipedes of arrests.
  • SUPPLY LINE: A point-and-click game about sorting school lunch items to “Make America Healthy Again.” Take bad food away from kids. That is the entire gameplay loop.
  • TRUMP SAVINGS TYCOON: A game where you fetch falling cash from entering a swimming pool—framed as filling your kid’s Trump Accounts.
wh-arcade-gov

A sixth card sits on the page labeled “Coming Soon” with the ominous tagline: “a new adventure awaits.” Nobody knows what that means yet, but given the pattern, your guess is as good as mine.

The Real Head-Scratcher Here

Beyond the obvious partisan messaging baked into every title, there is the question of execution. Multiple observers have pointed out that the entire arcade’s intro sequence bears a striking resemblance to the classic Xbox 360 startup animation.

Not “inspired by.” Not “homage.” It looks like someone ripped the assets, changed a few colors, and called it federal work product. The word “vibecoded” has been floating around heavily in reaction threads, and honestly, it is charitable.

This is not a leaked internal memo or a rogue staffer’s GitHub repo. This is sitting on an official .gov domain, promoted by the official White House social media account, and apparently created with some level of institutional blessing.

Whether this counts as a creative political statement, a legitimate educational outreach effort, or an expensive embarrassment depends entirely on your tolerance for taxpayer-funded game jams.

One thing is for certain. The discourse around this is going to be absolutely feral for the next several days.

Nice Try, Sony

Syndicate Editor
-
0
Nice Try, Sony

Sign up with the following links and enjoy the first 6 months of OpManager Nexus free.
On-Site: https://lmg.gg/site24x7
Cloud: https://lmg.gg/manageengine

NEWS SOURCES: https://lmg.gg/RVfnc

► SHOP OUR PRODUCTS: https://lttstore.com
► GET EXCLUSIVE CONTENT ON FLOATPLANE: https://lmg.gg/lttfloatplane
► LISTEN TO THE GAMING NEWS: https://lmg.gg/GLPodcast
► GET A VPN: https://lmg.gg/piagamelinked
► SPONSORS, AFFILIATES, AND PARTNERS: https://lmg.gg/partners
► WHERE WE BUY GAMES: https://lmg.gg/shopgames

——————————————————————-
Timestamps:
0:00 Sony’s State of Play and angry customers
2:17 GTA 6 hype, leaks, and lawsuits
4:52 QUICK BITS INTRO
5:06 Nexus Mods acquires SteamDB
5:51 CS2 slur tracker
6:41 Activision goes after cheat sellers
7:29 Resident Evil is too scary for Hideki Kamiya
8:23 A cow plays EVE Online
9:13 Credits

Go to Source

Standalone remove.bg to Close on December 1 Following Canva Integration

Mark Fowler
-
0
Standalone remove.bg to Close on December 1 Following Canva Integration

Remember when the internet first discovered remove.bg and it felt like absolute wizardry? Long before every mobile operating system and cheap photo app had built-in cutout tools, this simple site did exactly one task flawlessly: it stripped image backgrounds in a single click.

There was no bloated signup process and no massive software suite to load. Sadly, that era of standalone web utility is drawing to a close.

According to an official customer notification sent out by the team, the standalone remove.bg website will officially shut down on December 1, 2026, at 9:00 a.m. Central European Time (CET).

The platform’s entire background-erasing backend is being fully migrated into its parent company, Canva, which originally acquired the Austrian startup back in 2021.

remove-bg-home

This is a classic case of corporate consolidation, and frankly, it is a bit of a bummer. Instead of maintaining a lightweight, accessible tool, Canva is folding the tech into its own ecosystem to drive user acquisitions.

It is a logical business move, especially following Canva’s massive buyout of the generative platform Leonardo.ai, which is where the developer-facing API endpoints are currently being redirected.

If you are a regular user of the site, you need to audit your account immediately.

The shutdown timeline leaves no room for procrastination, and several systems will simply stop working overnight:

  • Burn Your Credits: Any unused Pay-As-You-Go (PAYG) credits sitting in your account will expire on the December 1 deadline. There are no refunds, so use them now or lose them.
  • Export Your Data: You will lose access to your dashboard history. You must manually download your credit usage reports and past financial invoices before the site goes dark.
  • Migrate Your API Keys: The self-service API will stop accepting requests on December 1. Developers must rewrite their integrations to point toward Leonardo.ai to avoid broken app workflows.
canva-background-home

It is hard to argue with the business logic here. Canva wants to build a unified design powerhouse to compete with Adobe, and keeping a legacy standalone site alive costs engineering resources.

But the loss of simple, single-purpose web tools is a frustrating trend. It forces consumers to sign up for massive, subscription-heavy ecosystems just to perform basic photo edits.

For those who do not want to buy into the broader Canva ecosystem, this closure will send them hunting for new independent alternatives.

It is the end of an era for one of the early web-based AI darlings, proving once again that in the current tech landscape, the big suites eventually swallow everything.

Clipto Secures $15M to Launch Local AI Video Search Apps and MCP for Agents 

Sato Kamoto
-
0
Clipto Secures $15M to Launch Local AI Video Search Apps and MCP for Agents 

SAN FRANCISCO, CA — August 31, 2026 — Generative AI is fantastic at spitting out high-quality video from simple text prompts, but it remains incredibly bad at figuring out what you did with the raw footage you shot three years ago. Creators and professionals are drowning in terabytes of unstructured video sitting completely dead on external drives.

According to the announcement, San Francisco-based startup Clipto just secured $15 million in fresh funding at a $250 million post-money valuation to fix this specific headache.

The cash injection, backed by global investment heavyweights like HSG (formerly Sequoia China), GL Ventures, and EnvisionX Capital, is aimed at expanding what Clipto calls the “memory layer” of the AI era.

Instead of forcing users to upload massive raw files to a slow, expensive cloud database, the platform uses proprietary on-device AI to run semantic video and audio searches, transcription, and summaries locally.

It is a smart pivot away from cloud-everything infrastructure.

clipto-homepage

Along with the funding, the team is shipping native Windows and Android apps, finally matching its existing iOS, Mac, and web offerings. But the real nerd-bait in this update is the introduction of Model Context Protocol (MCP) support.

This integration essentially gives autonomous AI agents a secure doorway to search and query a user’s private media library. If you are using digital assistants to automate video production workflows, they can now theoretically scan your local b-roll archives down to the exact timecode without leaking sensitive raw data to an external server.

While the concept of local semantic video search sounds a bit niche, the platform’s adoption metrics tell a different story. Clipto claims a footprint of over 30 million users, with enterprise adoption creeping into corporate teams at McKinsey, Apple, and Google.

Media outreach and strategic communication for the launch are being supported by Activate PR to target broader creator economies.

Frankly, the shift toward local, privacy-first AI memory is a necessary pushback against the bandwidth bottlenecks of current AI pipelines.

The real test will be whether Clipto can maintain its independent lead as operating system giants inevitably try to bake similar semantic media-indexing tools directly into the next generation of desktop hardware.

It’s Ternus Time

Syndicate Editor
-
0
It’s Ternus Time

Sign up with the following links and enjoy the first 6 months of OpManager Nexus free.
On-Site: https://lmg.gg/site24x7
Cloud: https://lmg.gg/manageengine

NEWS SOURCES: https://lmg.gg/z0zge

► SHOP OUR PRODUCTS: https://lttstore.com
► GET EXCLUSIVE CONTENT ON FLOATPLANE: https://lmg.gg/lttfloatplane
► GET A VPN: https://www.piavpn.com/TechLinked
► LISTEN TO THE TECH NEWS: https://lmg.gg/TLPodcast
► DIVE DEEPER ON THE LTT LABS WEBSITE: https://lmg.gg/labs
► SPONSORS, AFFILIATES, AND PARTNERS: https://lmg.gg/partners
► OUR PODCAST GEAR: https://lmg.gg/podcastgear

—————————————————
Timestamps:
0:00 John Ternus takes over as Apple CEO
1:16 GoPro merges with an optical photonics firm
2:15 Dyson’s $500 camera toothbrush
4:21 QUICK BITS INTRO
4:29 NYC restricts AI in schools
5:03 Acer Predator Atlas 7 and folding concept
5:30 Firefox gets ad blocking on iOS
5:49 Roku OLED TVs
6:20 KAIST hidden camera detector
6:53 Credits

Go to Source

Google Ad Manager Shifting Display Impressions to Begin-to-Render in February 2027

Sato Kamoto
-
0
Google Ad Manager Shifting Display Impressions to Begin-to-Render in February 2027

Google is quietly fixing a decades-old quirk in its advertising plumbing.

In an email notification sent out to Google Ad Manager network partners, the tech giant announced it will soon abandon its long-standing “count-on-download” (COD) method for display ads. Instead, Google is finally migrating to the industry-standard “begin-to-render” (BTR) methodology.

The shift itself is scheduled to take effect on February 17, 2027. Frankly, I believe this change is incredibly overdue. However, it is almost certainly going to cause some mild panic among yield optimization and ad operations teams when their dashboard numbers suddenly start to slide.

When you stop counting things that never actually showed up on a screen, the baseline numbers are bound to change.

google-ad-manager-keyart
Photo by Zulfugar Karimov on Unsplash

So, why does this distinction actually matter? Let us break down the technical differences between how Google registers your traffic today versus how they will handle it in 2027:

  • Count-on-download (The current state): Ad Manager counts a display impression the moment the ad server responds to a request and the creative asset starts downloading to a user’s device. It does not matter if the user immediately scrolls away, closes the tab, or has a terrible connection that kills the page load before the ad is visible.
  • Begin-to-render (The 2027 standard): In accordance with Interactive Advertising Bureau (IAB) and Media Rating Council (MRC) guidelines, an impression is only logged once the creative successfully loads and starts visibly rendering on the actual physical user device.

What does this mean for publishers? Well, expect your total reported impressions to take a hit. Any ad that began downloading but was abandoned before rendering will vanish from your reports.

While that might sound like a penalty, it actually aligns publisher metrics with buy-side Demand-Side Platforms (DSPs).

It should, in theory, reduce discrepancy disputes and lower administrative friction.

The update will apply broadly to display inventory across the web, mobile web, and Connected TV (CTV) banner environments. No action is required from network admins; Google is flipping the switch automatically.

If your ad ops team spends too much time manually reconciling differences between Google and third-party buyers, this change should fix that specific headache.

You can find more detail in the Google Ad Manager Help Center as the rollout date approaches.

JBL Launches Cove Whole-Home Wi-Fi Audio System Starting at $229

Sato Kamoto
-
0
JBL Launches Cove Whole-Home Wi-Fi Audio System Starting at $229

STAMFORD, CT, September 1, 2026 — If you have ever spent ten minutes screaming at your phone trying to get two wireless speakers in different rooms to play the same podcast in sync, you already know the smart home audio market is deeply annoying.

Most systems require constant regrouping, dropping connections the second you take a phone call. To fix this headache, HARMAN’s iconic brand is taking a direct swing at the whole-home audio establishment with the launch of JBL Cove, a new trio of Wi-Fi-connected speakers built for zero-friction listening.

The company is launching three distinct models to choose from, each aimed at slightly different use cases around the house. Unlike traditional Bluetooth speakers that pair directly to a single device, these Wi-Fi-enabled boxes stream content directly from the cloud. This means your playlist continues uninterrupted even if your phone leaves the house or rings.

The lineup drops on September 13, 2026, starting at $229.95 for the compact M1, jumping to $399.95 for the portable P1, and topping out at $449.95 for the flagship X1 system.

imageye___-_JBL-Cove-FAMILY-Living_Room_Lifestyle_0709_x5

The engineering behind these units focuses heavily on automated acoustic correction. Each speaker in the Cove family utilizes proprietary Constant Sound Field technology alongside automatic self-tuning and AI Sound Boost.

This hardware-software mix constantly measures the boundaries of your room, adjusting the output on the fly to minimize distortion when you decide to crank the volume during a weekend gathering.

The mid-tier P1 and flagship X1 also feature up-firing drivers to support spatial Dolby Atmos audio.

According to the official release announcement, JBL is banking on an “app-optional” user experience to win over buyers tired of crowded phone screens. While you can use the JBL One app to customize EQ levels, a physical “Moment” button on the physical chassis lets you trigger features like Spotify Tap or activate “Music Follows Me.”

This feature literally transfers whatever you are listening to from one room’s speaker to another with a single physical press—no smartphone required.

FeatureJBL Cove M1JBL Cove P1JBL Cove X1
How it fits in the homeCompact home Wi-Fi speakerPortable home Wi-Fi speaker3D room-filling home Wi-Fi speaker
Drivers2 x 30mm and 1 x 20mm silk dome neodymium tweeters & 1 x 4” woofer3 x 30mm and 1 x 20mm silk dome neodymium tweeters (1 up-firing) & 1 x 4” woofer4 x 30mm silk dome neodymium tweeters (1 up-firing) & 2 x 4” woofers
Dolby Atmos®NoYesYes
Battery—Up to 14 hours: just drop it on the dock to recharge—
IP rating—IP56—
Auto room calibrationYesYesYes
Constant Sound FieldYesYesYes
AI Sound BoostYesYesYes
Stereo PairYesYesYes
Whole-home audioYesYesYes
Works with JBL Bar MK2YesYesYes – for multi-room music only
Price$229.95$399.95$449.95
AvailabilitySeptember 13, 2026September 13, 2026September 13, 2026
* JBL Bar 300/500/700/1000/1300XMK2 Series
** JBL Cove P1 and Cove M1 at launch
***Alexa+ is available in the US to Prime members and Alexa+ subscribers.

What makes this ecosystem particularly clever is how it integrates with existing living room hardware.

If you own a compatible JBL Bar Series MK2 soundbar, you can wirelessly bind the Cove P1 or M1 speakers to act as dedicated rear surround-sound channels.

This dual-use capability is a smart move, giving buyers a clear path to upgrade from simple multi-room background music into a full-scale home theater setup over time.

JBL_Logo

There is also an interesting tech-partnership layer here.

Alongside standard features like AirPlay 2, Google Cast, and Spotify Connect, the Cove lineup comes with built-in access to Alexa+, Amazon’s upgraded, subscription-based artificial intelligence assistant.

While legacy Alexa handles basic smart home tasks for free, the inclusion of Alexa+ shows that JBL is positioning these speakers as smart home hubs first, and simple music players second.

Whether customers will actually pay a monthly fee to talk to their speaker is still up in the air, but the acoustic hardware itself looks incredibly solid.

Inside the Picometer Tech Built for NASA’s Next Flagship Space Telescope

Mark Fowler
-
0
Inside the Picometer Tech Built for NASA’s Next Flagship Space Telescope

BROOMFIELD, CO, August 30, 2026 — Everyone loves the razor-sharp, ultra-narrow snapshots coming out of James Webb, but deep space research actually requires massive wide-angle sweeps to make sense of cosmic evolution.

NASA launched its latest flagship observatory today—the Nancy Grace Roman Space Telescope—off a pad at Kennedy Space Center in Florida. It is a major win for wide-field astronomy.

Behind the scenes, aerospace giant BAE Systems designed and built the Opto-Mechanical Assembly for the Wide Field Instrument (WFI), which acts as Roman’s primary scientific engine.

The hardware keeps the optical bench, thermal controls, and precision electronics locked in place under extreme conditions so the sensors can capture huge slices of the sky without distortion.

Roman-launch-image-8-30-26
NASA’s Nancy Grace Roman Space Telescope launching from Kennedy Space Center (BAE Systems).

So what makes this hardware actually matter? The WFI gives Roman a field of view at least 100 times wider than Hubble while surveying deep space up to 1,000 times faster.

Frankly, I think this wide-net strategy is what will finally give astrophysicists a realistic head count on exoplanet demographics, rather than just picking off isolated star systems one by one.

According to a press release from BAE Systems, Bonnie Patterson, vice president and general manager of Civil Space, noted that Roman will track how billions of planets, stars, and galaxies assemble across deep time.

The mission builds directly on the engineering lessons learned from both Hubble and Webb, continuing a long line of contractor support for NASA’s heavy-lift space science.

Engineers are not waiting around for Roman’s raw data to start flowing before moving to the next project, either. BAE is already deep into early design work for NASA’s next planned flagship, the Habitable Worlds Observatory (HWO). That upcoming mission aims to directly image Earth-like planets orbiting distant stars to search for atmospheric signatures of life.

Under its Ultra-Stable Large Telescope Research and Analysis (ULTRA) contract, BAE is testing mirror actuation systems capable of tweaking optics at the picometer scale. That is a trillionth of a meter—a level of mechanical control necessary if you want to freeze out light from a parent star and isolate the faint glow of an orbiting world.

Big flagship space hardware takes decades of unglamorous mechanical engineering before anyone gets to look at a processed image file.

Roman is finally sitting in space, but the real work starts when those massive data pipelines begin dumping raw cosmic surveys back down to ground stations on Earth.

Serious Sam: Shatterverse Drops on PC and Consoles with 5-Player Co-Op Action

Alex Quake
-
0
Serious Sam: Shatterverse Drops on PC and Consoles with 5-Player Co-Op Action

AUSTIN, TX, August 31, 2026 — Long before modern shooters got bogged down in hyper-realistic tactical movement and complex battle passes, Serious Sam was all about one thing: holding down the left mouse button while backpedaling away from hundreds of screaming headless kamikazes.

Devolver Digital and Behaviour Interactive are trying to bring that pure, chaotic boomer shooter energy into the modern era with the launch of Serious Sam: Shatterverse on PC, Xbox Series X|S, and PlayStation 5.

Priced at an aggressive $19.99—with a 10 percent discount for early adopters during launch week—this is not just another standard level-by-level sequel.

Developed by Behaviour Interactive, the studio best known for Dead by Daylight, Shatterverse takes the franchise’s signature horde-clearing gunplay and wraps it inside a co-op, procedurally shifted roguelite structure.

serious-sam-shatterverse-keyart

Multiversal Madness and 5-Player Squads

The story kicks off when Sam’s arch-nemesis, Mental, fractures the dimensional boundaries, splintering the universe into the “Shatterverse.” To fix it, players must team up with alternate-dimension versions of Sam stone-cold style.

For the first time in series history, up to five players can jump into a lobby together to blast through shifting environments and take down Mental’s five lieutenants. If you prefer working alone, you can also run the entire campaign solo.

According to an official release update on Steam, Shatterverse blends hand-crafted arenas with randomized run modifiers, hidden portals, and high-risk anomalies. As you progress, you unlock permanent upgrades and wild boons that supercharge your arsenal, transforming familiar weaponry into completely irresponsible tools of mass destruction.

A Modern Twist on Classic Horde Mechanics

The core gameplay features designed to keep players looping through runs include:

  • Expanded Arsenal: A mix of classic Serious Sam heavy guns and completely new weapons plucked from alternate dimensions.
  • Procedural Runs: Hand-crafted arenas that shift conditions, enemy compositions, and risk opportunities every single run.
  • Meta-Progression: Earn permanent upgrades to expand run possibilities and unlock game-bending modifiers over time.
  • Classic & New Foes: Massive hordes featuring iconic franchise monsters alongside brand-new enemy types under Mental’s command.
serious-sam-shatterverse-gameplay

“Reviving the legendary gunplay, massive enemy hordes, and unforgettable bosses that Serious Sam is known for, while blasting into bold new roguelike territory was challenging, but extremely rewarding,” said Behaviour Interactive’s Nic Duchesne in the announcement.

He noted that the project was heavily guided by community feedback to ensure it delivers a complete, social experience at launch.

Boomer shooters have enjoyed a massive revival over the past few years, but adding procedural roguelite mechanics can be a double-edged sword. If Behaviour Interactive managed to hit the right balance between frantic skill and progression, Shatterverse might just be the shot in the arm this classic franchise needed.

1...456...182Page 5 of 182