I've been running a benchmark for comparing generative image models with a strong emphasis on complex prompt adherence since the days of SDXL back in 2023.
It’s not really intended to be viewed chronologically but there is a progress metric that shows how models have improved their scores over time, which might help you compare earlier models with later models such as Ideogram4 or Moss-Image released in 2026.
If they already have a health specific model why aren't the others instructed to forward health related queries? Sounds to me, they are just trying to cover their ass; "Your honor, it is not our fault she didn't use the correct model"
It's especially bad if you're non-American with multiple TVs.
e.g. I've got three screens in my house (office TV, living room TV, basement projector) in Canada — upgrading to the latest Apple TV on each of them would cost me $1,287.
I have two televisions - living room TV with an AppleTV and bedroom TV with some Roku device. I'm going to buy the new AppleTV and put it in the living room, then move that AppleTV to the bedroom, then recycle the Roku (finally, I hate that thing).
It should be less expensive but I rely on it (cancelled cable TV and just have streaming now) and so I'm just happy they are still working on it.
Depending on what you subscribe to, streaming services are now as expensive (if not more) than cable was if you got premium channels. I hear also Disney will add ads to all tiers as well.
Neural Viz had really good consistency and really good scripts, also importantly knew to write the story around the limitations in what AI could do, including using non-human characters to avoid that uncanny valley that I immediately see in the Kono Express video. Truly a master at his craft, it's a shame he hung it up!
No, Cerebras is serving GPT-5.6-Sol (released in July).
OpenAI is back to using Nvidia chips (with small batch sizes) for the ultrafast mode of their latest models.
That takes the speed down from from 750tok/s on Cerebras to 300tok/s, and presumably is much more expensive than the Cerebras solution, but Cerebras has supply constraints. I believe they are planning aggressive data center expansion in this quarter but much harder to scale up than Nvidia chips.
It's still a pretty new field but Jev handily beats Luna-decisions on the Artificial Analysis Decision Index on every metric (price, latency, and accuracy)
I have a custom wrapper for 1PW that lets an agent browse and request credentials, then I can provide it a lease to access those for a certain amount of time. Super nice and I hope they build that in to 1PW itself.
I do this too and have a wrapper around tool calls which masks the password on tool call output, and unmasks on input. So the LLM just sees a temp credential
Just an idea, is it possible to automatically rotate secrets immediately after they're used or are these things like passwords to online platforms that need to go through a reset flow.
No, not as far as I can figure. Sure, Claude can access secrets with the CLI but it's not very granular usually it's just asking for access to all secrets
Given the lack of repercussions for AI companies that have hacked various services, there will be an AI company that trains their medical agents on PHI hacked from Epic/Cerner and gets away with it.
It shows his comp over a number of years, including these enormous 3 year packages that are a ceiling, not a guarantee.
Picked 2025 specifically instead of the 2022 three year package or the 2026 three year package, to make it more comparable to what you might expect from a pre-IPO company like Anthropic; otherwise it would be too much apples and oranges.
In some ways I prefer looking at these over the hyper-polished 2026 image generators tbh.
reply