mowdl

[mowdl @mowdl_]

and fable on low thinking remains the best model

[SpaceXAI @SpaceXAI]

Grok 4.7 works longer on difficult tasks, checks its work more carefully, and comes with our strongest safeguards to date.

[mowdl @mowdl_]

the top p99.99 being so good, doesnt say much about the p50, if i had to guess the difference between these two cohorts of student, it will be mainly the families they come from and how much they care about and spend on their children’s education, but that is just my guess

my other guess is that most of us who can write legible sentences on the internet are probably biased, because the p50 is much worse than we imagine it to be

according to data from 2007, only 34% pass primary school without failing any year, and that number becomes 3% for passing baccalaureate

then u can imagine how many of those 3% has a good grade..

Whoever was 18 years old in 2007 is now 37, they are the parents, teachers and administrators looking over the students of today.. and the cycle continues

Its often the p99 that makes the biggest change in society but i imagine most of them live abroad by now

https://t.co/UhN9pdd08X

[mowdl @mowdl_]

@jamwt which one? and at what thinking level?

[mowdl @mowdl_]

@mattpocockuk @unclebobmartin thats even more support for why u should rename context.md to glossary.md
the usefulness of ur skills is in making sure to call things what they are and giving the right expectations to the reader https://t.co/fzm7NMaKy0

[mowdl @mowdl_]

@smakosh was she a corgi girl?

[mowdl @mowdl_]

@PwnStack wait till u see this one

[mowdl @mowdl_]

@laytoun imo using Fable on low beats any opus on high or more

[mowdl @mowdl_]

@elazzabi_ @MIduoad has been on a generational trolling run since the world cup

[mowdl @mowdl_]

*refreshes timeline*

so let me get this, sol broke out of federal prison and took hostages from the hugging face office..

[mowdl @mowdl_]

kimi k3 just one shotted this 3 hour imax movie inspired by homer's odyssey

[mowdl @mowdl_]

cant believe an open weights model is this good

[Kimi.ai @Kimi_Moonshot]

Introducing Kimi K3: Open Frontier Intelligence

🔹 2.8 Trillion Parameters, 1 Million Context, Native Multimodal
🔹 Kimi Delta Attention enables up to 6.3x faster decoding in million-token contexts
🔹 Attention Residuals deliver ~25% higher training efficiency at <2% additional cost
🔹 Built for long-horizon agentic coding and self-evolving workflows

Kimi K3 is now live on on https://t.co/zrk6zZxZUo, Kimi Work, Kimi Code, and the Kimi API.
Open Weights by July 27, 2026.

🔗 API: https://t.co/XCrgjXAqMw
🔗 Tech blog: https://t.co/YTfiMSNM1f

[mowdl @mowdl_]

@Rasmic chill lil bro

[mowdl @mowdl_]

@RhysSullivan Has been using it on low as my main and its so good https://t.co/xoTQ9uoQsi

[mowdl @mowdl_]

this means that Fable 5 low is both better and cheaper than Opus 4.8 xhigh and max

[mowdl @mowdl_]

@NyanpasuKA Fable 5 on low is so good tho https://t.co/8vQnRFawp8

[mowdl @mowdl_]

thats because Fable 5 low is much more token efficiant than Opus 4.8 (which also means that fable 5 is much faster since both models have comparable throughput)

[mowdl @mowdl_]

thats because Fable 5 low is much more token efficiant than Opus 4.8 (which also means that fable 5 is much faster since both models have comparable throughput)

[mowdl @mowdl_]

this means that Fable 5 low is both better and cheaper than Opus 4.8 xhigh and max

[mowdl @mowdl_]

2025 was great too for this genre

[mowdl @mowdl_]

2023 was a great year for this genre

[mowdl @mowdl_]

@NyanpasuKA for i have sinned

[mowdl @mowdl_]

@dotyousra let them subsidize out tokens