After a slight blunder with an American retailer, Sonos has officially made its two new speakers official. 

Alongside the Sonos Play, which is positioned as one of the brand’s most flexible speakers, is the Sonos Era 100 SL. But how does the Era 100 SL compare to the similarly titled Era 100? 

Considering we awarded the Sonos Era 100 a 4.5-star rating, why should you buy the Era 100 SL?

To help you decide, we’ve compared the specs of the Sonos Era 100 SL to the Era 100 and pointed out the key differences and noteworthy similarities between the two below. Keep reading to see which speaker is likely to suit you best.

Not sold on either? Visit our list of the best Bluetooth speakers and best outdoor speakers to find your new favourite model.

Advertisement

Price and Availability

At the time of writing, the Sonos Era 100 SL is available for pre-order and will officially launch in stores on March 31st. With an RRP of $189/£169, it’s hailed as being Sonos’ “most cost-effective way to build” a home setup.

Instead, the Sonos Era 100 has an RRP of £199/$219. However, as it’s an older speaker, it is possible to find the device with a price cut, so make sure you keep an eye out during sales events.

Advertisement

Sonos Era 100 supports voice control

One of the biggest differences between the Sonos Era 100 SL and Era 100 is that the latter has a microphone and, therefore, supports voice control. You can either use Sonos Voice Control to manage music, or you can also rely on Amazon Alexa for answering questions, setting timers and the like. 

Advertisement

On the other hand, the Sonos Era 100 SL isn’t equipped with any voice control or assistants, as it lacks a microphone. Sonos explains that the decision to not equip the Era 100 SL with built-in voice control was to ensure the speaker is more affordable than the 100. However, if you pair the Era 100 SL with a voice-enabled soundbar then you’ll still benefit from voice control, even though the Era 100 SL doesn’t actually support it.

Android users won’t be able to use Trueplay tuning technology on the Sonos Era 100 SL

Following on from the above, the Era 100 features Sonos’ clever Trueplay technology which measures how sound reflects off walls and furnishings in a room and tunes the speaker accordingly. While the Sonos Era 100 SL does also support Trueplay, it will only work with an iOS device. This is because the Era 100 SL doesn’t have a microphone, which Android devices need to activate Trueplay. 

With this in mind, if you’re sporting one of the best Android phones and want optimised sound for your room, then the Sonos Era 100 SL might not be the top choice for you. 

Both have the same sonic architecture

Under their respective hoods, the Sonos Era 100 SL and Era 100 sport the same sonic architecture. In fact, Sonos promises that the more affordable Era 100 SL will have “no compromises on sound quality”. 

Advertisement

Considering we concluded that the Era 100’s provides a weightier performance across the frequency range, this certainly bodes well for the Era 100 SL. We also found the Era 100 sported an increase in bass over its predecessor that gave music genres such as R&B, Hip Hop and Pop more impact and excitement too.

Advertisement

Having said that, we’d encourage users to play around with the EQ to perfect the bass and treble levels to best suit your personal listening preferences. Although we haven’t reviewed the Era 100 SL yet, we expect this will be the case for the new speaker too.

Sonos Era 100 SL is designed to fit into your existing speaker set-up

Although you can listen to the Era 100 SL on its own, Sonos explains it’s designed to slot into your existing home entertainment set-up. For example, you can pair two Era 100 SLs to a Sonos soundbar, such as the Sonos Beam or Sonos Arc Ultra, to create a surround sound system.

Alternatively, Sonos does state that you can use the Era 100 SL as a stand-alone speaker too.

Advertisement

It’s a similar situation with the Era 100, which can be paired with another Era 100 SL and a soundbar too.

Advertisement

Both share similar designs, although the Era 100 SL is slightly lighter

At first glance, you’d be forgiven for thinking the Era 100 and Era 100 SL were the same speakers, as both measure in at 7.19 x 4.72 x 5.14 inches and come in a choice of Black or White. However, unsurprisingly as it lacks a microphone, the Era 100 SL weighs slightly less at 1.95kg compared to 2.02kg. 

Early Verdict

As we haven’t reviewed the Sonos Era 100 SL yet, we’ll hold off from giving a conclusive verdict. However, as it’s slightly more affordable than the Era 100 and promises the same brilliant sound quality and sleek design, the Era 100 SL is certainly an appealing option for many – as long as you’re an iOS user and won’t miss voice control.

Advertisement

Every GPU cluster has dead time. Training jobs finish, workloads shift and hardware sits dark while power and cooling costs keep running. For neocloud operators, those empty cycles are lost margin.

The obvious workaround is spot GPU markets — renting spare capacity to whoever needs it. But spot instances mean the cloud vendor is still the one doing the renting, and engineers buying that capacity are still paying for raw compute with no inference stack attached.

FriendliAI’s answer is different: run inference directly on the unused hardware, optimize for token throughput, and split the revenue with the operator. FriendliAI was founded by Byung-Gon Chun, the researcher whose paper on continuous batching became foundational to vLLM, the open source inference engine used across most production deployments today.

Chun spent over a decade as a professor at Seoul National University studying efficient execution of machine learning models at scale. That research produced a paper called Orca, which introduced continuous batching. The technique processes inference requests dynamically rather than waiting to fill a fixed batch before executing. It is now industry standard and is the core mechanism inside vLLM.

This week, FriendliAI is launching a new platform called InferenceSense. Just as publishers use Google AdSense to monetize unsold ad inventory, neocloud operators can use InferenceSense to fill unused GPU cycles with paid AI inference workloads and collect a share of the token revenue. The operator’s own jobs always take priority — the moment a scheduler reclaims a GPU, InferenceSense yields.

“What we are providing is that instead of letting GPUs be idle, by running inferences they can monetize those idle GPUs,” Chun told VentureBeat.

How a Seoul National University lab built the engine inside vLLM

Chun founded FriendliAI in 2021, before most of the industry had shifted attention from training to inference. The company’s primary product is a dedicated inference endpoint service for AI startups and enterprises running open-weight models. FriendliAI also appears as a deployment option on Hugging Face alongside Azure, AWS and GCP, and currently supports more than 500,000 open-weight models from the platform.

InferenceSense now extends that inference engine to the capacity problem GPU operators face between workloads.

How it works

InferenceSense runs on top of Kubernetes, which most neocloud operators are already using for resource orchestration. An operator allocates a pool of GPUs to a Kubernetes cluster managed by FriendliAI — declaring which nodes are available and under what conditions they can be reclaimed. Idle detection runs through Kubernetes itself.

“We have our own orchestrator that runs on the GPUs of these neocloud — or just cloud — vendors,” Chun said. “We definitely take advantage of Kubernetes, but the software running on top is a really highly optimized inference stack.”

When GPUs are unused, InferenceSense spins up isolated containers serving paid inference workloads on open-weight models including DeepSeek, Qwen, Kimi, GLM and MiniMax. When the operator’s scheduler needs hardware back, the inference workloads are preempted and GPUs are returned. FriendliAI says the handoff happens within seconds.

Demand is aggregated through FriendliAI’s direct clients and through inference aggregators like OpenRouter. The operator supplies the capacity; FriendliAI handles the demand pipeline, model optimization and serving stack. There are no upfront fees and no minimum commitments. A real-time dashboard shows operators which models are running, tokens being processed and revenue accrued.

Why token throughput beats raw capacity rental

Spot GPU markets from providers like CoreWeave, Lambda Labs and RunPod involve the cloud vendor renting out its own hardware to a third party. InferenceSense runs on hardware the neocloud operator already owns, with the operator defining which nodes participate and setting scheduling agreements with FriendliAI in advance. The distinction matters: spot markets monetize capacity, InferenceSense monetizes tokens.

Token throughput per GPU-hour determines how much InferenceSense can actually earn during unused windows. FriendliAI claims its engine delivers two to three times the throughput of a standard vLLM deployment, though Chun notes the figure varies by workload type.

Most competing inference stacks are built on Python-based open source frameworks. FriendliAI’s engine is written in C++ and uses custom GPU kernels rather than Nvidia’s cuDNN library. The company has built its own model representation layer for partitioning and executing models across hardware, with its own implementations of speculative decoding, quantization and KV-cache management.

Since FriendliAI’s engine processes more tokens per GPU-hour than a standard vLLM stack, operators should generate more revenue per unused cycle than they could by standing up their own inference service. 

What AI engineers evaluating inference costs should watch

For AI engineers evaluating where to run inference workloads, the neocloud versus hyperscaler decision has typically come down to price and availability.

InferenceSense adds a new consideration: if neoclouds can monetize idle capacity through inference, they have more economic incentive to keep token prices competitive.

That is not a reason to change infrastructure decisions today — it is still early. But engineers tracking total inference cost should watch whether neocloud adoption of platforms like InferenceSense puts downward pressure on API pricing for models like DeepSeek and Qwen over the next 12 months.

“When we have more efficient suppliers, the overall cost will go down,” Chun said. “With InferenceSense we can contribute to making those models cheaper.”

Amazon’s AI assistant Alexa+ is getting another new personality. On Thursday, the company announced it’s expanding its lineup of personality styles for users to choose from to include a “Sassy” option, which is for adults only. Notes Amazon, before opting to use the Sassy personality, users will be required to go through additional security checks in the Alexa app.

The personality style will also not be available when Amazon Kids is enabled, Amazon says.

The new option joins others like Brief, Chill, and Sweet, launched last month.

When you toggle on the option for Sassy in the Alexa mobile app, you’re warned that the Sassy style uses explicit language, which is why it requires a security check. On iOS, this involved a Face ID scan.

The AI assistant explained its style to us like this: “The Sassy style is built on one premise: help first, judge always. Every answer comes wrapped in wit and a well-placed roast — it’ll answer your question; it’ll just make you feel something about it first. Expect reality checks delivered with charm, compliments that somehow sting, and warmth you didn’t see coming. Equal-opportunity irreverence, zero apologies. Honest, sharp, and funny — and somehow that’s more helpful than helpful.”

Alexa’s app also had warned that the style could contain “mature subject matter.”

However, further investigation discovered this is not Amazon’s version of something like Grok’s adult AI companions. The AI assistant said the new option won’t get into areas like explicit sexual content, hate speech, illegal activities, personal attacks, or anything that could cause harm to oneself or others.

The move is the latest example of how Amazon is trying to make Alexa+ more customizable, as it revamps the assistant for the generative AI era. By offering the assistant different personalities — including one positioned as more adult — Amazon is borrowing from a broader trend in AI, where companies have been experimenting with tone, style, and personas to make their assistants more engaging and personalized to the individual users’ choices.

from the the-truth-business dept

We’ve already all seen what the Ellison family’s version of “editorial independence” looks like over at CBS, where contrarian troll Bari Weiss has turned the already very Republican friendly news giant into a safe space for right wing zealots and autocrats. All overseen by a Brendan Carr chosen censor tasked with ensuring the channel always makes Donald Trump happy.

As always with authoritarian regimes (and corporate ownership), this is all presented to the public as an effort to restore balance, eliminate (nonexistent) “liberal bias,” and reach out to real Americans. As if billionaires and their useful idiots could care less about everyday Americana.

After being gifted two Hollywood studios and two major news empires by daddy and Donald Trump, fail-upward nepobaby David Ellison made the rounds last week to insist that CNN’s “editorial independence” would be retained under Paramount/CBS ownership. His evidence? CBS:

“So, look, I’ve said this since the beginning, which is, you know, for — when it really comes to — editorial independence will absolutely be maintained. It’s maintained at CBS. It’ll be maintained at CNN. And, really, who we want to talk to is the 70% of Americans and really around the world that identify as center-left, as center-right. And we want to be in the truth business. We want to be in the trust business. And that’s not going to change.”

Of course, if anybody had actually been paying attention to CBS, they’d see how the network under Weiss has already tried to repeatedly kill stories that aren’t favorable to Donald Trump, gone out of its way to normalize right wing opportunists like Erika Kirk, and has driven away a lot of remaining CBS journalists with Weiss’ obvious efforts to pander to Trump and Netanyahu.

Like CBS, CNN already goes well out of its way to be extra friendly to authoritarians. The network has routinely faced criticism for consistently airing sneering MAGA devotee Scott Jennings. Under Ellison ownership there’s zero serious doubt, by anyone, that CNN will become even more friendly to autocrats. After they get done firing untold thousands of people to try and pay down the deal’s immense debt.

Traditionally there’s only one editorial direction U.S. journalism usually goes under consolidated corporate ownership. U.S. media owners like tax cuts, deregulation, subsidies, access, and merger approvals, so corporate media’s editorial slant generally follows the financial interests of ownership. The pretense that U.S. media suffers from widespread “liberal bias,” or the belief that there are still functional firewalls between ownership and editorial, are long-deceased relics.

Larry Ellison clearly wants to hoover up what’s left of corporate media (including CBS, CNN, HBO) — and fuse it with his co-ownership of TikTok to create a sort of Hungary-esque autocratic state media, where administration allies praise dear leader while the government strangles independent and public media just out of frame.

The only thing saving us from the full and terrible vision of this outcome to date is the fact that very few of the weird nepobabies and brunchlords being tasked with its creation have anything you’d mistake for competence.

Filed Under: authoritarian, bari weiss, consolidation, david ellison, journalism, layoffs, media, propaganda, state media

Companies: cbs, cnn, paramount, warner bros. discovery

When Max Brodeur-Urbas co-founded Gumloop in mid-2023, his vision was to help non-technical employees automate repetitive tasks using AI. At that time, the concept of AI agents was still largely experimental and prone to errors.

As AI technology has matured, so has Gumloop’s offering.

The company claims that it now allows teams at organizations like Shopify, Ramp, Gusto, Samsara, Instacart, and Opendoor to deploy reliable AI agents that autonomously handle complex, multi-step tasks, all without ever needing an engineer.

Employees can share the agents they build with colleagues, creating a compounding effect that accelerates internal automation. “They get addicted, they start building more agents, and then all of a sudden, the whole company is AI native,” Brodeur-Urbas told TechCrunch.

As companies race to adopt AI, Benchmark general partner Everett Randle believes the key to success lies in empowering every worker with AI superpowers, and Gumloop’s intuitive agent-builder is an example of the kind of tool that will unlock that potential.

That’s why Randle, who joined Benchmark last October from Kleiner Perkins, chose to lead a $50 million Series B investment into Gumloop. The deal, which is Randle’s first at his new firm, included participation from Nexus VP, First Round Capital, Y Combinator, Box Group, The Cannon Project, and Shopify.

Though Gumloop wasn’t actively seeking new capital, the startup decided this was the year to “step on the gas.” For Brodeur-Urbas, partnering with Benchmark—the firm behind icons like eBay, Uber, and Dropbox—was a “no-brainer.”

While Brodeur-Urbas previously planned to ‘build a 10-person, billion-dollar company,’ the surging demand from enterprise clients has compelled him to build a dedicated sales force and scale up his engineering team, he said.

Gumloop is by no means the only player vying to turn every knowledge worker into an AI agent-builder. The startup faces stiff competition from established automation platforms like Zapier and n8n, as well as specialized agent builders like Dust. Even foundational AI labs are entering the fray.  For instance, Anthropic’s Claude Co-Work allows users to create autonomous agents without writing a single line of code.

But Randle believes Gumloop is superior to all its rivals. During his due diligence, he discovered that at least one of the company’s customers had adopted Gumloop somewhat organically.

When Randle asked a CTO how they chose Gumloop, the response was telling. The company had given employees full access to Gumloop alongside two competitors. Six months later, the results were clear: staff were using Gumloop daily or weekly, while the competing tools sat untouched, Randle told TechCrunch.

The reason Gumloop gained such momentum, according to Randle, is its minimal learning curve. “You can go in and start making agents and workflow automations immediately,” he said.

While many AI startups worry that foundational models will replicate the same functionality and render them obsolete, Randle is convinced that Gumloop’s model-agnostic approach is precisely what will keep attracting customers.

As models continue to evolve, one may perform better than another for a specific task. So, Gumloop provides the flexibility to choose the model best suited for the job at any given moment.

Another reason why model independence is attractive, according to Randle, is cost. “Plenty of enterprises have OpenAI, Gemini, and Anthropic credits. They want to use all of them,” he said

His excitement for the company ultimately comes down to the sheer size of the opportunity.

“Enterprise automation is a massive pot of gold,” Randle said. “I think it’s the biggest category in enterprise AI.”

Ever since 3D printing has become a popular tool, the question of waste has been looming in the background. The sad reality of rapid prototyping is that you’re going to generate a lot of prints that just don’t aren’t fit for purpose, even if your printer runs them off perfectly every time. Creality has some products on the way aimed at solving that problem, and [Embrace Making] on YouTube has got his hands on a pre-production prototype of the Creality M1 Filament Maker to give the community a first look.

The M1 is actually only half of the system; Creality is also working on an R1 shredder to reduce your prints into re-usable shreds. [Embrace Making] hasn’t gotten his hands on that, but shredding prints isn’t the hard part. We’ve featured plenty of DIY shredders in the past. Extruding filament reliably at home has traditionally proven much more difficult, which is why we mostly outsource it to professionals.

Lacking the matching shredder, and wanting to give the M1 the fairest possible shake, [Embrace] tests the machine out first using Creality-supplied PLA pellets. The filament diameter isn’t as stable as we’ve gotten used to, and the spool rolling setup needs a bit more work.

Again, this is an early prototype. Creality says they’re working on it and claims they’ll get to ±0.05 mm precision in the production models. Doubtless they’ll also fix the errors that led to [Embrace]’s messy spool. That’s probably just software given that the winding mechanism did a pretty good job on the Creality-supplied spool.

Most importantly, the M1-produced filament does print. The prints aren’t perfect due to the variation in diameter, but they turn out surprisingly well for home-made filament. [Embrace] also shows off the ability to mix custom colors and gradients, but, again, using raw PLA rather than shredded material. Hopefully Creality lets him test drive the R1 shredder once its design is further along.

This is hardly the first time we’ve seen a filament extruder. The goal of this product is to pair with a shredder and use it for recycling, but if you’re going to stick with raw plastic pellets, you may as well print them directly.