Many Tesla owners are fed up with Elon Musk’s political fools and find an modern way to show their dissatisfaction.

Some owners decided to scrape Tesla’s “T” brand on their cars to protest against the controversial CEO activities. Since Musk liked President Donald Trump, and one other place in his administration, the leftist owners of Tesla decided to distance themselves.

Photos of cars that appear to be Teslas with various logos have turn out to be viral. Vehicles with the body of some of its high -class models now have Branding Audi, Honda or Toyota. attached. Even provocative, but popular Cybertrucks gained a brand new name stuck to them.

For some, having Tesla not promotes pride, but more “embarrassing” supporting Muska. Many considered the last straw to use the alleged Nazi salut by Musk during the inauguration of Trump. Especially for European owners Tesla, these “strange” activities They broke their loyalty To Tesla.

“He became more and more strange,” said Mike Schwede, longtime owners of Tesla based in Switzerland. “I felt nothing but complete disgust. And I didn’t like to sit in my Tesla anymore. “

People not only change from Teslas, which they have already got, but they aren’t intended to buy recent ones.

Data obtained by the European Car Manufacturers Association revealed that Tesla sales in Europe fell by 45%in January. Tryget from Texas sold lower than 10,000 cars, 9945 specifically in a month, despite the sale of over 18,000 a yr earlier.

What’s more, Musk’s last branding around the continent associated him with fascist leaders than at a technological innovator focused on the environment. The Tesla Reclon parody in London is now ridiculed by the fascist Salute Muska, also referring to the vehicle in “Swasticar”. In addition, a number of “anti-elon” stickers appeared on EVS, including people who say: “I bought it before Elon went crazy.”

Because Musk supports the far -right politicians and movements in the United States and a number of other European countries, the liberal database of consumers of his company takes matters in his hand. While Musk stays aware of the decreasing sales, he encouraged his employees that we were a “epic” yr before us.

AI models are thrown out at a shocking pace by everyone, from large technology corporations, akin to Google, to startups akin to OpenAI and Anthropic. Tracking the newest could be overwhelming.

The confusion is that AI models are sometimes promoted based on industry comparative tests. But these technical indicators often do not reveal much how real people and corporations actually use them.

To cut the noise, TechCrunch has developed a review of probably the most advanced AI models released since 2024, with the main points of their use and what they are for. We will update this list concerning the latest premieres.

There are actually over 1,000,000 AI models: for instance, hugging your face, It runs over 1.4 million. So this list can miss some models that work higher in a method or one other.

AI models released in 2025

Aya Vision Cohere

Cohere has released a multimodal model called Aya Vision, which, he claims, is the perfect in the category in doing things like signatures of images and answering questions on photos. He also cooperates in languages ​​apart from English, unlike other models. Is available to For free at whatsapp.

Openai’s GPT 4.5 “Orion”

OpenAi calls Orion the biggest model to date, promoting its strong “world knowledge” and “emotional intelligence”. However, worse ends in some comparison tests compared to newer reasoning models. Orion is on the market for subscribers of the OPENAI $ 200 plan monthly.

Claude Sonnet 3.7

Anthropic says that that is the primary “hybrid” reasoning within the industry, because it will probably each decelerate quick answers and really take into consideration all the things if needed. It also gives users control over how long the model can think, on anthropic. Sonnet 3.7 is on the market to all Claude users, but heavier users will need a Pro 20 USD plan monthly.

Xia’s Gook 3

GROK 3 is the newest flagship model recognized by Elon Musk Startup XAI. His claimed Ahead other leading models of mathematics, exact sciences and coding. The model requires x Premium (i.e. USD 50 monthly). After one study found Grok 2 leaned left, musk promised Move the grocs more “politically neutral”, however it shouldn’t be yet clear if it has been achieved.

OpenAI O3-Mini

This is the newest model of OPENENAI and is optimized for tasks related to STEM, akin to coding, mathematics and science. His Not the strongest OpenAI model, but since it is smaller, the corporate says This is a much lower cost. It is on the market at no cost, but requires subscription for heavy users.

Deep Research Openai

Deep Openai research has been intended for in -depth research on the topic with clear quotes. This service is barely available with the Pro Chatgpt 200 USD subscription monthly. Openai It recommends In the case of all the things, from learning to purchasing research, but think that hallucinations remain an issue for AI.

Mistral Kot

Mistral has launched Le Chat applications, Multimodal AI assistant. Mistral claims Le Chat answers faster than some other chatbot. It also has a paid version with Current journalism from AFP. Tests from the world An impressive le chat performance was found, although he made more mistakes than chatgpt.

Openai operator

The OpenAI operator is to be a private intern that may do things independently, for instance, help buying food. Requires a CHATGPT Pro subscription price $ 200 monthly. AI agents have many guarantees, but they are still experimental: Washington Post reviewer says the operator He decided to order a dozen eggs for $ 31, paid by the reviewer’s bank card.

Google Gemini 2.0 Pro Experimental

The very anticipated flagship model Google Gemini says that it’s distinguished by coding and understanding of general knowledge. It also has a really long context window of two million tokens, helping users who’ve to quickly process huge fragments of the text. The service requires (at the least) Google One AI Premium subscription in the quantity of $ 19.99 monthly.

AI models released in 2024

Deepseek R1

This Chinese model AI stormed the Silicon Valley. R1 Deepseek works well in coding and mathematics, while his nature open source implies that everyone can start it locally. Besides, it’s free. However, R1 integrates the censorship of the Chinese government and is within the face of growing bans to potentially send user data back to China.

Gemini Geed Reactione

Deep research summarizes Google search ends in an easy and well -cited document. Service It is useful for college kids and everyone who needs a fast summary of the research. However, its quality shouldn’t be pretty much as good as real paper paper. Deep research requires Google One AI Premium subscription in the quantity of $ 19.99.

The goals are called 3.3 70b

This is the newest and most advanced version of AI Open Source Meta Open Source. The finish advertised this version As the most cost effective and best, especially due to mathematics, general knowledge and instructions. It is free and open source.

Openai Sora

Sora is a model that creates realistic movies based on the text. Although it will probably generate whole scenes, not only clips, Openai admits that he often generates “unrealistic physics”. Currently, it is barely available in paid chatgpt versions, ranging from Plus, i.e. USD 20 monthly.

Alibaba Qwen QWQ-32B-PP-Review

This model is one among the few who compete with O1 Openai in some industry comparative tests, excellent in mathematics and coding. Ironically, for the “model of reasoning”, it has a “place to improve reason for reason”, Alibaba says. He also utters the censorship of the Chinese government, shows TechCrunch tests. It is free and open source.

Use of anthropic computer

The Claude computer goals to take control of the pc to perform tasks, akin to coding or booking a flight ticket, which makes it the predecessor of the OpenAI operator. However, the use of a pc, Remains within the beta version. Prices happen via API: USD 0.80 for million input tokens and USD 4 for million production tokens.

X.Ai’s GROK 2

AI Elon Musk, X.Ai, launched an improved version of its flagship GROK 2 Chatbot It claims It is “three times faster”. Free users are limited to 10 questions every two hours per grok, while premium and premium subscribers+ plans+ have higher use limits. X.Ai also launched the Aurora image generator, which produces highly photorealistic images, including some graphic or violent content.

OpenAI O1

The OPenai O1 family is aimed toward providing higher answers through “thinking” through answers through a hidden reasoning function. The model leads in coding, mathematics and safety, Openai claimsBut he also has problems with cheating people. Using O1 requires chatgpt plus subscription, which is $ 20 monthly.

Anthropic’s Claude Sonnet 3.5

Claude Sonnet 3.5 is model anthropic claims as the perfect within the classroom. He has turn into known for the probabilities of coding and is taken into account a technological informant Chatbot by selection. The model could be obtained at no cost at Claude, although heavy users will need a monthly PRO subscription in the quantity of USD 20. Although he can understand images, he cannot generate them.

Openai GPT 4-Mini

Opeli advertised 4-mini GPT as probably the most inexpensive and fastest model, thanks to its small size. Meaning To enable a big selection of tasks akin to supplying customer support chatbot. The model is on the market at a free chatgpt level. It is healthier suitable for easy tasks with large volume compared to more complex ones.

Cohehe Command R+

Coher Command model r+ Excels in complex generation (or RAG) applications within the scope of recovery (or RAG) for enterprises. This implies that it will probably really find and quote specific information. (RAG inventor It actually works in Cohere.) Mimo to Rag nie rozwiązuje w pełni problemu halucynacji AI.

Despite the clouds of uncertainty approaching the economy and geopolitics, people still wish to travel. To satisfy this demand, MewsRecognized as Amsterdam “Unicorn” constructing the SaaS platform for hotel management, he collected a brand new $ 75 million.

Financing appears at a crossroads within the tourism industry. On the one hand, they’re positive signals. Analysts in Oxford Economics It forecasts that the industry will increase by 12% this 12 months, working to 1.75 billion “arrivals”.

And the travel budget corresponds to money for travel startups. Travelperk collected $ 200 million in January with a valuation of $ 2.7 billion; Hostaway raised $ 365 million in December 2024; And Tourlane in Germany raised $ 26 million from Sequoia in November 2024. Prosus also raised his despense from Latin American America for $ 1.7 billion.

But that is not all pink. It shouldn’t be clear how tariffs and other economic maneuvers will affect people’s travel budgets, not to say uncertainty about geopolitics. It can be value noting that Expedia was control of staff -Could or not it’s a one -time move to enhance performance, or is it more to come back?

Mews also has dozens of competitors, including siphouses reminiscent of Oracle, in addition to other startups, reminiscent of cloudbed supported by Softbank.

For now, things appear to hum for Mews.

Tiger Global, a brand new supporter, led this latest round, with the participation of previous Kinnevik investors, batteries and growth capital in Alternaties Goldman Sachs. (The battery led the B MEWS series $ 33 million, while Kinnevik and Goldman Sachs ran the C series of $ 185 million in 2022; Kinnevik also ran a series D 2024 in the quantity of $ 110 million).

This last round is opportunistic, said general director Matt Welle in an interview with TechCrunch. In particular, it appeared after Tiger approached Mews who attempt to cooperate more. Mews doesn’t reveal the present valuation, and Welle as an alternative called it the general round of height, adding that the startup plans to lift a much larger round throughout the 12 months or two.

For a certain context: exactly a 12 months ago, when Mews, who was recognized as Amsterdam, raised his D series, he was valued at $ 1.2 billion.

Mews is growing. Currently, about 6,300 hotels use the platform, in comparison with 5000 a 12 months ago, and said that in 2024 he processed payments value $ 10 billion, twice as much as a 12 months ago. In 2024, his revenues exceeded $ 200 million because he added more customers to the platform. It also extends services on this platform, which also builds greater customer revenues. Some are ecological, and a few in response to mergers and acquisitions, reminiscent of the recent takeover of Atomize to assist users manage revenues. Notes from Pitchbook Mews also raised $ 100 million in debt in September last 12 months to finance acquisitions.

(*75*) a big a part of the MEWS business is in Europe (he says that it now has 20% of the market penetration within the roof region), a big increase comes from lots of pressure through the pond.

Nowadays, the founder (and co-coo) Richard Valt spends most of its time within the US, where the corporate has concentrated very much-it has entered North America within the last 12 months.

When Valt founded the corporate in 2012, he understood many pain points related to conducting hotel activities. He himself was an independent owner of the hotel and realized that there was a most important opportunity to make use of the Internet and increase software as a service to construct something recent.

There are many points on the hotel market, but Mews accepts the “toast” approach to the industry, constructing its product with the concept of being end-to-end. Offers hotel availability and online booking software, directly and via other corporations; Hotel personnel management tools and facilities; Payment and accounting tools; Event management software; Loyalty and guests applications and more.

Perhaps predictably, Welle said that Mews had recently doubled AI. It has an internal team working on various projects-not only agency tools addressed to the client, but additionally algorithms to enhance the operation of MEWS. He said that one example is a brand new option of the client’s profile that adopts all of the actions and preferences of individual individuals who register within the hotel chain to create a summary of “Tweets size”, which could be used by conjers and receptionists to enhance customer support.

“Mews again defines what it means to ensure unique experiences of guests in the hotel and gastronomic sector,” said Sara Eadie in an announcement, who conducted the investment in Tiger Global. (Tiger refused to interview this story).

As he developed, Mews focused on catching larger fish within the space of hospitality. The company claims that the variety of “brands” of hospitality that use its tools has doubled, and a few of its recent customers are Best Western Hotels, Weender and Lark Hotels.

To make it clear, this doesn’t mean that each one Best Westerns now use Mews. As with all contracts with clients who run the franchise model, MEWS is now on the approved list of suppliers, but still must conclude a contract with individual franchises to win a brand new company.