Ramp launches its own AI model router, called Router
9:46 AM PDT · August 20, 2026
Corporate expense management platform Ramp is hot on the heels of Stripe in setting up toll houses for AI inference.
Ramp on Wednesday evening launched its own AI model routing service, dubbed Router, that lets users and companies use and switch between various large language models through an API. The company says it has been using the router it built for its own AI usage needs over the past three years.
The service is only available in the United States. It’s free to use for the remainder of 2026 (users will still have to pay for AI model inference costs), and it comes with a $26 credit launch offer. The company did not say how much the service will cost next year.
In its function, Router is pretty similar to how OpenRouter operates, though the latter offers many more AI model options than Ramp’s current offerings.
Router offers access to models from OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai. It also provides several “strategies” to help its customers route AI requests to models based on their preferences. For example, one lets users set a preference for model providers’ flex usage tiers, while another lets Router choose which model to route queries to based on up to three user-specified benchmarks. Users can also choose to route only difficult problems to expensive models or test models easily without having to switch.
Users get a dashboard, too, that lets them see token spend, cost, latency, fallback attempts, and other details.
Notably, Router has an opt-out data retention policy: it will record model inputs, outputs and tool calls for one year by default, though the company says it will remove “personally identifiable information before using that content to improve the product.”
For Ramp, entering the model routing business offers a two-pronged opportunity: it gets to tap the booming AI inference market, and also offer its existing clients a model routing service that fits in neatly with its existing products, which includes AI token usage monitoring and token spend management.
And, if Router proves as attractive of a model testing arena as OpenRouter has, Ramp may also be able to build long-standing relationships with AI labs and inference providers worldwide. That could help Ramp, which raised $750 million at a $44 billion valuation in June, gain new customers and a new point of entry for selling its expense management products.
Meta brings Pocket, an app that lets you vibe-code and share games, to US users
9:07 AM PDT · August 20, 2026
Meta’s experimental vibe-coding gaming app, Pocket, is now rolling out to everyone in the U.S. The app, which arrived quietly last month in the test market of Brazil, allows people to generate small, interactive games using AI prompts, which are published to a scrollable feed.
Based on Meta’s acqui-hire of the team at the vibe-coded gaming platform Gizmo earlier this year, these games — or “gizmos” as they’re called — respond to touch and the tilt of your phone, play sound effects, and can include clips of your favorite songs.
They can also use photos from your camera roll or access your camera. The resulting games can then be shared on your profile, where others can save them, remix them into other creations, or simply repost them.
The app is the latest example of Meta’s push to make AI creation tools mainstream, following its earlier efforts that included making AI-generated images with its Meta AI app as well as AI videos created with an experimental app called Vibes.
Pocket now joins these and several other standalone mobile applications from Meta in recent months. CEO Mark Zuckerberg has credited the increased output to AI-enabled software development, which makes it faster for the company to test and ship new ideas.
“Earlier this year, we shipped Instagram Instants. We also just launched Forum, a stand-alone Groups app, and Seller, a stand-alone Marketplace app. I expect it to become a lot easier to ship new apps,” he told analysts on July’s earnings call. “So we are planning to build out more ideas and use our recommendation systems to scale them.”
In addition to Instants, Forum, and the Seller app, Meta has shipped an experiment involving AI bedtime stories, and this week, a Meta AI app for Mac.
The 2026 Startup Battlefield 200 is here — see who made the cut
9:00 AM PDT · August 20, 2026
We’re thrilled to unveil the 2026 Startup Battlefield 200 — TechCrunch’s annual, handpicked list of the 200 most promising early-stage startups from around the world.
These founders rose to the top from thousands of applications spanning every industry and geography. Their reward? A coveted spot to exhibit at TechCrunch Disrupt 2026, taking place October 13–15 in San Francisco’s Moscone West. This is the cohort to watch — and this fall, you’ll get to meet them in person.
From AI to climate, health to fintech, robotics to consumer — these are the companies shaping the future across every major category.
A special thank-you to Google for Startups and SAP, our premier sponsors, for helping early-stage startups build, scale, and succeed.
Only at Disrupt can you:
- Watch live demos of never-before-seen tech.
- See Battlefield companies pitch in front of top global VCs.
- Cheer them on as they compete for $100,000, equity-free, and the iconic Disrupt Cup.
Are you a VC scouting the next breakout investment? A tech enthusiast looking to stay ahead of the curve? An early-stage founder who missed the Startup Battlefield 200 application deadline because you were too busy building a robot in your garage?
Step onto the battlefield. Join us at Disrupt 2026. Register for your ticket before prices go up in September.
Presenting the 2026 Startup Battlefield 200
Agtech and food
Muju Earth Technologies Limited NakedPak
Automotive and transportation
Laplace Intelligence Motion Sync Technologies Inc. MyMonthlyCar Inc. Neumo
Biotech and pharmaceuticals
Alithea Biotechnology Apertum Neuroscience BioTrillion
Consumer
Cybersecurity
E-commerce and retail
Edtech
Energy and clean tech
Enterprise tech
Entertainment and media
Fintech
Government, legal, and regulatory
Health and wellness
Human resources and recruiting
Logistics and supply chain
SSINDEX | Stakeholders Sustainability Index
Manufacturing and industrial
Marketing and advertising
Real estate and proptech
Smart cities and infrastructure
Space and defense
Every year, Startup Battlefield alumni go on to build category-defining companies — including Dropbox, Cloudflare, and Discord. With over $32 billion raised and 250 exits, the Startup Battlefield track record speaks for itself.
The top 20 finalists from this list will be revealed live on Day 1 of Disrupt on October 13, so be sure to follow along.
If your startup made the list, keep an eye on your inbox — we’ll be in touch shortly with time-sensitive next steps. If you applied but don’t see your company here, you’ll also receive an official update from us soon. Due to the high volume of submissions, we’re unable to respond to individual application inquiries.
As for the rest of you …
Get out of your inbox and get ahead of the game at TechCrunch Disrupt 2026 on October 13-15 at Moscone West in San Francisco.
Secure your spot at TechCrunch Disrupt. Prices will go up in September.
Inertia Enterprises finds a way to make its fusion fuel fast
9:00 AM PDT · August 20, 2026
Fusion startup Inertia Enterprises said it has found a way to cut the time it takes to make its fuel pellets from several days to just minutes.
By slashing fuel filling time, Inertia says it has knocked down one of the 10 barriers it must overcome to deliver the first phase of its commercial power plant ambitions. The startup gave TechCrunch an exclusive first look at the process.
The advance was not guaranteed: Investors gave Inertia $450 million on the premise that it could commercialize technology developed at the National Ignition Facility (NIF), which is an elaborate science experiment that requires painstaking care and feeding to operate. At the NIF, making fuel pellets can take a week or more and cost a small fortune. That’s not a great recipe for a profitable operation.
“But when you really double-click on it, you’re like, wait a minute, they’re only making a handful of them a year,” Jeff Lawson, co-founder and CEO of Inertia, told TechCrunch. “I put on my commercial hat and was like, wait a minute, I know the word for this: Prototypes.”
At Inertia, the team set about turning those prototypes into something mass-manufacturable. “That’s why we’re hiring people from the likes of Apple,” he said. “There’s a bunch of industrial engineers who are like, ‘Alright, I guess I have to go figure out how to make that a billion times over in a factory.’”
The NIF’s fuel pellet is far from an iPhone, though. The outside is a spherical diamond shell, and just inside, there’s a thin layer of frozen deuterium and tritium, the isotopes of hydrogen that can fuel a fusion reactor. Inside the crystalline layer lies a mix of gaseous deuterium and tritium. Each solid layer must be as close to perfectly spherical as possible.
Those fuel pellets are then wrapped in gold casings known as hohlraums, which convert laser energy into X-rays that compress the fuel pellet inside. If everything goes to plan, that compression causes atoms to fuse and release energy. But even small imperfections in the spherical shape can disrupt the ignition process, preventing a fusion reaction from reaching its full potential.
Inertia had to condense the process to the point where it makes commercial sense without straying too far from the physics discovered at the NIF.
“What happens if you try to do it faster?” Lawson said. The team had a head start: Annie Kritcher, co-founder and chief scientist at Inertia, designed the first fusion experiment at the NIF that released more power than it consumed.
After several rounds of development, the startup was able to grow the crystals in about 30 minutes, something that could take up to a week at the NIF. Altogether, a single fuel pellet can be made in about two to three hours, and the process can be ramped to industrial scale. Inertia developed the process with help from the NIF at the Lawrence Livermore National Lab, with which the startup has formed a public-private partnership.
Inertia has an advantage that the NIF does not, though. Because the startup is planning to use a laser that is four times more powerful than the one currently at the NIF, it can tolerate more imperfections in the fuel pellet. This also helped speed the manufacturing process.
“We actually have a lot of margin,” Lawson said. “That’s our strategy, to oversize our driver, our laser, to give us lots of margin to go play with in every other part of the system.”
Reducing fuel filling time has the knock-on effect of reducing the amount of tritium Inertia needs to hold at any given time. Tritium is radioactive, and handling it requires care. It’s also extremely expensive at the moment, about $30,000 per gram, and only about 25 kilograms are stockpiled globally, according to the journal Science.
Inertia, like many fusion startups, plans to make its own tritium with help from the fusion reactions, but it still needs some inventory to get started. Plus, reducing manufacturing time helps keep inventory small. Inertia expects its full-scale commercial power plant will use 10 fuel pellets every second.
“By reducing the latency of this step, you’ve made your facility smaller; you’ve made the whole thing faster, the whole thing more efficient,” Lawson said.
