Software Developer and Entrepreneur | Building https://xitter.cf/t.co/k5nu8N3iea - Local AI For Mac
Silicon Valley
Joined April 2024
- Tweets1.9K
- Following202
- Followers92
- Likes23.3K
Ash retweeted
YAXEL LENDEBORG GOES OFF TO HELP WARRIORS GET THEIR FIRST WIN OF THE SEASON 😤🔥
31 PTS
12 REB
6 AST
Ash retweeted
There’s ongoing work to improve efficiency and cache efficiency that should allow more tasks to be completed using Argon within quota.
Ash retweeted
Yaxel Lendeborg on getting support from his vets after last game and Luka Doncic calling him a “24-year-old rookie”:
“I’m always gonna be myself, man. I don’t care what anybody has to say about it. The guys are over there on the couch talking bad about me like, come on, bro, I could care less what you got to say. What else do they say? The 24-year-old rookie thing? I’m 24, in the NBA, with the best job in the world, just moved my mom out to California. Like, I’m living the dream. I could care less, you know what I mean? So I’m always gonna be myself. That’s the way I was taught, the way I was raised. I mean, you know, I’m happy with life no matter what goes on.”
(via @joseph_dycus)
Ash retweeted
Draymond Green says if you don’t rock with a guy like Yaxel Lendeborg that who competes like that, who willing to fight like that, you’re probably a loser:
“If you don’t respect that, you are probably a loser. A soft loser that when someone gets on you, you can’t take it. You probably have a victim mindset and everything that falls in that category because if you can’t get behind that all the people like, “Oh, man he shouldn’t be talking why shouldn’t he?” Why shouldn’t he be talking? That’s like this old adage of like, oh you a rookie you don’t talk, fu*k that why wouldn’t I talk? And that’s been my mindset from day one coming into this league.”
Ash retweeted
Hey @Xbox, I don't appreciate my background being turned into an unchangeable ad for a game.
Readers added context they thought people might want to know
The background is not unchangeable; disable "Show event-themed backgrounds" in Settings > General > Personalization > My background to restore the prior one.
windowscentral.com/gaming/call-of…
purexbox.com/guides/modern-…
Ash retweeted
Sam Altman: By the definition most people had five years ago, we’ve already surpassed AGI.
“These models are smart now, and they’ll keep getting smarter.”
“More and more people will think we’ve gotten to an AGI system every year.”
“We have systems now that are really increasing people’s productivity and doing valuable economic work.”
“A system capable of autonomously discovering new science would feel almost definitively like superintelligence to me.”
“And I think that would be a wonderful thing for the world.”
The Artificial Analysis Benchmark should not be used as a serious measure of model capability at this point
They literally had to change their entire benchmarking system because GPT 5.6 Sol and Astra had the same intelligence score in their original system
It’s all made up
Kimi K3.1 Pro launches at 4th on the Artificial Analysis Intelligence Index scoring 55, making large gains in agentic performance; it will be the leading open-weights model once its weights are published
@Kimi_Moonshot's new reasoning model gains 11 points over Kimi K3 (max, 44). It trails only Claude Opus 5.5 (max with fallback, 58), Claude Sonnet 5.5 (max with fallback, 56), and Muse Wonder Full (max, 56) on the Intelligence Index.
Once the weights for K3.1 Pro are released it will be the highest-scoring open weights model and the gap between the leading open weights and proprietary models on the Intelligence Index will narrow from 12 points to 3.
Key results:
➤ Large gains in agentic performance: Kimi K3.1 Pro scores 3rd on GDPval-AA with 1,846 Elo, up 312 Elo from Kimi K3, behind only Claude Opus 5.5 (1,866 Elo) and Muse Wonder Full (1,857 Elo). Kimi K3.1 Pro also achieves the highest score of any model on Terminal-Bench 4.0. K3.1 Pro scores 67%, up from 13% for Kimi K3 and ahead of Claude Sonnet 5.5 (64%). AutomationBench-AA, which tests agentic SaaS workflows, rises from 58% to 73%.
➤ Gains in other evals are more moderate: Humanity's Last Exam rises from 47% to 52%, CritPt from 23% to 25%, and SciCode from 59% to 61%. AA-Omniscience Index rises from 20 to 30, as the hallucination rate falls from 53% to 44% and accuracy rises from 48% to 51%.
➤ Cost per Task roughly equal to Kimi K3 at $1.98 per Task: K3.1 Pro uses 55K output tokens per Intelligence Index task, 14% more than Kimi K3 at 48K, while its first-party token prices are 33% lower.
Additional model details:
➤ Context window: 1M tokens.
➤ Size: 2.8T total, 104B active parameters
➤ Input modalities: Text and image
➤ Weights and license: Weights currently not published, Moonshot is expected to release the weights soon after launch
➤ Pricing: $2/$10/$0.20 per 1M input/output/cache hit tokens, 33% lower than Kimi K3 on all three rates.
Congratulations to the team at @Kimi_Moonshot on the release!
I have used Meta Muse and OpenAI Dot and Meta Muse is much better
With dot I have had to take over its computer countless times
I have not had to take over Muse’s machine once
Ash retweeted
A manosphere podcaster who urged men to leave their wives because they can "do better" would be (correctly) utterly reviled.
The double standards need to end.
