Prediction: KYC will kill distillation. Chinese models will no longer be able to compete. This will further increase the gap between Anthropic/ OpenAI and the rest. LLMs are not commoditizing.
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.
Access to all other Claude models is not affected.
We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.
Read our full statement: https://t.co/bwn0sximKZ
@kunchenguid It creates lock-in. The models are increasingly fine tuned to their harnesses which increases switching cost. E.g., the model has a certain expectation of the tools in the harness are supposed to work.
Really don’t understand the tech obsession with trying to automate the process of booking flights… I think I travel reasonably frequently and it’s really not a big deal to click around for a couple of mins to book a flight…
@thsottiaux I think it’s not a good choice to show scheduled automation runs as normal chats. I have an automation that runs every hour and it fills up my chat sidebar. Feel like it should be in the automations page
Biggest mistake I see VCs making as a technical CEO: investing in things that seem defensible that don’t make sense.
For example, fine tuning models is idiotic for most vertical companies, but VCs thought it sounded defensible. Now a wave of fine tuned model companies are dying.
@kimmonismus Couldn’t agree more. Already the first interview triggered me so much where they said that it was irresponsible of him to release a product like this due to cybersecurity concerns. Madness
The bitter lesson for web agents
The last 1 year has taught us a new bitter lesson that we think others are not yet grokking.
Agents that look at the web like humans (using screenshots of sites) navigate and generalize better than agents that read code (HTML, DOM).