• The Myth of Fable
  • Fox's Roku Move
  • Snap's Newest Spectacle

The Myth of Fable
After months of intentional delay, Anthropic finally released its public follow-up to its infamous Mythos model, the model so capable that it could not be released to the unwashed masses. Fable 5 is/was Anthropic's attempt to provide the power of Mythos with the guardrails necessary to protect all of us from ourselves. It came with a 1 million token context window for long and complex instructions and a maximum capacity of 128,000 output tokens. But with much power came great cost: Fable 5 is priced at double the cost of Claude Opus models with an MSRP of $10 per million input tokens and $50 per million output tokens. Just as everyone was beginning to assess the impact of the price increase while OpenAI pondered price decreases, along came the U.S. Department of Commerce with an unexpected left-hand jab - an export ban on Fable 5 and Mythos 5. This ban meant that no foreign nationals could have access to these models, even if they worked for Anthropic and had helped in creating them. There is probably nothing quite like being told you no longer have access to a thing you helped build. With no way to limit access to meet the Commerce Department's demands, Anthropic simply shut down the models for everyone.

Much has been written about the role Amazon played in this decision and even more has been written about the already problematic relationship between the US government and the senior leaders at Anthropic. It's probably nearly universally understood at this point that the two groups are not on the best of terms. Perhaps the tenuous ties play into the government's actions and subsequent resistance to negotiate with Anthropic on potential solutions. Is there in fact a solution that could be deemed workable by both sides? It seems impossible to tell.

Anthropic had been on a bit of a tear prior to releasing Fable 5; its products are growing in popularity with large companies and regular people alike. It had theoretically overtaken OpenAI in valuation and had started its initial public offering process in advance of OpenAI. Things were looking up for Anthropic and at times there even seemed to be tiny slivers of hope of defrosting the ice that had formed around their relationship with the Trump administration. But this latest move seems almost as if the Trump administration decided that things needed to be slowed down and Anthropic needed to be taken down a notch. Of course, this may be just perception, but the fact remains that this development puts a big speed bump on Anthropic's road to go public and establish themselves as a profitable AI powerhouse. And somewhere cocooned in all the layers of mistrust and drama lies the fact that a potentially valuable model is no longer available to anyone, anywhere. Entities, including nation states that have had friendly relationships with the US, are unable to take advantage of either Mythos or Fable to try to advance their own cybersecurity capabilities or just take advantage of new and more productive workflows. And with the already inexplicable situation of having banned Anthropic's products for governmental use while simultaneously using Anthropic's products for governmental use, the US has further complicated its own path forward for artificial intelligence supremacy. If the concern is that powerful models are a risk to national security, then there is cause for alarm among all big tech companies producing frontier models. But if the concern is specific to Anthropic, it will become more obvious and more difficult to defend as time passes, further complicating relationships and adding more fuel to the fire of lawsuits. And for Anthropic, a company that seems to struggle to communicate forthrightly and clearly with the US government (or really, anyone), this may be the pivotal moment where the story shifts to a competitor and the momentum Anthropic had is lost.

Fox's Roku Move
On Sunday, I was discussing with my husband the best possible scenarios for Roku if they were serious about being bought. He suggested that Google would be a good suitor because it would allow them to gain more viewer data and establish deeper relationships with TV manufacturers. I wasn't completely convinced this was the best choice, but before I even had time to sit and think about it, we learned that Fox had jumped in with both feet. This also did not seem initially like the best choice to me, but when framed as an attempt to compete better with Netflix and Amazon, it made it bit more sense. I don't think of Fox as a competitor in that space and that's probably mainly because it wasn't. Owning Roku gives them more surface space for advertising and cross-marketing, and in that regard, might help narrow the gap with those competitors. Fox currently owns ad-supported streaming service Tubi and combining Tubi with the Roku channel will reach a larger segment of the free TV-streaming audience. Tubi has about 1% of that audience and while Roku has only about 3%, it is the fastest growing free streaming service in the US. And with the data of more than 100 million households with Roku products, Fox will be able to increase its advertising reach and compete directly with the bigger players on that front. Fox will also have more eyeballs to present its products such as Fox Nation and Fox One to hopefully acquire more customers from Roku's world.

At the same time, there are some risks with this acquisition. Roku is currently a neutral player where streaming services can advertise and promote their products. Now those companies would theoretically need to pay a direct competitor for access to the Roku platform. Is the platform sticky enough and lucrative enough to continue that approach or would other platforms like those available on Samsung and LG TVs become more appealing? And will Fox stay true to its stated commitment to keep Roku open and partner-friendly or will it slowly degrade service over time in an effort to gain control or just better monetize the asset? And will manufacturers like Hisense and TCL that had been making Roku products accelerate their shift to other platforms?

Fox will need to navigate the relationship with customers, viewers, manufacturers, and competitors/partners very carefully to make this acquisition successful. Media merger history seems to suggest the road will be rocky.

Snap's Newest Spectacle
Look, I'll level with you: the internet does not seem to think Snap made the right move here. And what is that move, you may ask? Snap unveiled their latest face gadget, the Snap Specs, which many are calling the Apple Vision Pro in glasses form with no puck or dongle. The Specs are very chunky but somewhat fashionable glasses with full-color screens with 51 degrees of view. They have cameras for augmented reality functionality as well as capturing photos and video. They can do the types of things that we've seen other smart glasses do such as answer questions and provide guidance on what is visible in the field of view. They have dual Snapdragon processors and a proprietary display system that supports screen casting, virtual whiteboarding, and a developer ecosystem based on Claude Code, Codex, and Cursor. From a hardware perspective, they unite some of the best aspects of Meta's Display smart glasses and the many XR display glasses made by companies like XREAL, Viture, and TCL RayNeo. You can use the glasses to watch video, play games, talk to friends and family, and get work done. In terms of what they can do, they are the closest thing available to the Apple Vision Pro outside of the Apple ecosystem and they aren't that much more expensive than Samsung's Galaxy XR headset but are significantly smaller and lighter. When you view them in the context of the Vision Pro or Galaxy XR, they seem worlds ahead in their consumer viability. But when viewed from the perspective of traditional smart glasses like the Meta Ray-Bans, they look unnecessarily large and unwieldy. It is a difficult middle place in which to exist. For anyone willing to live in that space, they offer hand tracking, audio and video display and recording, phone notifications, and will supposedly offer up to 4 hours of battery life. They will be priced at $2195. This is undoubtedly a lot of money, but again, these seem to do more than the Galaxy XR and are more than $1000 cheaper than the Vision Pro while being much easier to wear. As a hardware geek and gadget girl, I think what Snap has done is very interesting and compelling for certain uses cases.

But at the same time, price and weight do matter and people are comparing these against other pairs of smart glasses, which isn't doing Snap any favors as most of those cost less than $800. Pricing them so high is a tough sell for most. It probably would have been better for Snap to take Meta's approach and offer cheaper and simpler smart glasses at a much lower price but still offer the Specs to appeal to price-insensitive early adopters and tech influencers (who could market them on social media). Going with just one product at such a high price will limit sales, even if the glasses are amazing and do well all the things Snap has suggested they can do. Nevertheless, I do think this push toward smaller stand-alone XR devices is the way the market must go in order for customers to warm up to the idea of computers on their heads and faces. Even if this round isn't successful for Snap, someone else is going to take up this mantle and eventually produce glasses that are small and light enough for people to be willing to try them and wear them in public. Hopefully Snap survives long enough to be part of that final story.