AI High Signal

1 curation · 626 membres · fenêtre de 24 h, demi-vie 6 h, seuil 3 relayeurs

Fenêtre 6 h 12 h 24 h 2 j 3 j 7 j
Montrer Les deux Liens 0 Tweets 9
Seuil 2 relayeurs 3 relayeurs 4 relayeurs 5 relayeurs 6 relayeurs Tout remettre
0liens
9tweets
35873posts lus
529auteurs
3239liens en base
1477 hde recul

Les tweets 9

Axe indépendant : repris sans regarder les liens.

1
Claude @claudeai · 09/11 23:11
Introducing Claude Fable 5.2 and Claude Mythos 5.2, designed to tackle the most daunting technical challenges. They’re capable collaborators on ambitious projects in many fields, and show improvements in areas like coding, reasoning, and vision.
15 relayeurs à 15 · il y a 5 h Voir sur X
Qui l'a repris
Andreas Kirsch 🇺🇦 elvis” elie” TomLikesRobots🤖 Daisy Hollman” swyx 🔜 aie.codes”
Ce qu'ils en ont dit
elvis @omarsar0 a cité · 09/11 21:50
Also, it looks like Anthropic doesn't want you to build RSI?! On a more serious note, even with Opus 5.5, kernel and LLM optimization tasks get refused most of the time. Open models are the only way!
elie @eliebakouch a cité · 09/11 21:18
with this, i think it's quite likely that anthropic will never publicly release (or at least not in the coming months) a model more capable than opus 5.5 at llm training
Daisy Hollman @TheWholeDaisy a cité · 09/11 20:50
I'm so excited that I finally get to talk about our new model with you all publicly! Let us know what you think, and show us all of the cool things you're building!
swyx 🔜 aie.codes @swyx a cité · 09/11 20:44
btw by complete coincidence the $ANT IPO roadshow across wall street started today. they had this in the chamber and its also nice that it came after Astra, Argon and Carbon…
2
D
David Andrews @broyojo42 · 04:20
I had Opus 5.5 formalize CPython in Lean: 613,318 lines of Lean that copy Python's internals line by line, tested against the real interpreter and checked by Lean. On the way it found real bugs in Python, including crashes from plain Python code. 🧵
3 relayeurs à 3 · il y a 13 min Voir sur X
Qui l'a repris
Terry Yue Zhuo Ofir Press Andreas Kirsch 🇺🇦
3
Alexandr Wang @alexandr_wang · 00:50
1/ announcing our smallest, most efficient little model is here: muse ember 🔥 built for work that agents repeat all day, it gets you more intelligence per dollar than Haiku, Luna, and all other flash models at any reasoning setting and does it at a third of the cost or less!
3 relayeurs à 3 · il y a 4 h Voir sur X
Qui l'a repris
Ananya Kumar Zach Mueller Jack Rae”
Ce qu'ils en ont dit
Jack Rae @jack_w_rae a cité · 09/11 19:50
Releasing a small and mighty model today, Muse Ember 🔥 The team really cooked on this one (👏)
4
H
hari_haran (building archgen YC F26) @HariAyapps · 00:09
@archgen_ is building skyscrapers for semiconductor chips Shrinking transistors is getting harder. Stacking them makes heat harder to manage. We’re building the tools to design chips in three dimensions. We stack transistors vertically, shorten connections, and optimize timing
3 relayeurs à 3 · il y a 4 h Voir sur X
Qui l'a repris
Krish Gupta” Austin Lyons” elvis”
Ce qu'ils en ont dit
Krish Gupta @krishgupta72 a cité · 00:09
Super bullish on the team!!
Austin Lyons @austinsemis a cité · 09/11 20:13
Kind of sounds like Tau scaling? @HariAyapps is this “traditional EDA” moving to 3D? Don’t incumbents account for 3D already? Also you didn’t mention LLMs, kind of surprised?
elvis @omarsar0 a cité · 09/11 18:47
Chip design is next for AI. Transistors are getting harder to shrink, so chips are going vertical. Stacking them shortens connections but makes heat and timing much harder to manage. Archgen is building the tools to design chips in three dimensions. Their AI agent reads each
5
DesignArena @DesignArena · 09/11 21:59
BREAKING: Muse Ember by @AIatMeta has redefined the Preference vs Speed Pareto frontier on Overall HTML, knocking out 10 top models from @GoogleDeepMind, @OpenAI, and @DeepSeek_ai. At 33s per generation, Muse Ember is 9.5× faster on average than the previous Pareto frontier
3 relayeurs à 3 · il y a 6 h Voir sur X
Qui l'a repris
Jack Rae Alexandr Wang” Grace Li”
Ce qu'ils en ont dit
Alexandr Wang @alexandr_wang a cité · 09/11 19:58
pareto frontier who? 🏎️ 💨 - muse ember 🔥
Grace Li @grx_xce a cité · 09/11 19:38
Huge congrats to Muse Ember by @Meta for setting the Pareto frontier on overall one-shot HTML!
6
W
William Shen @shenbokui · 09/11 20:43
Our first version of Muse Ember 🔥 Ember aims to maximize intelligence per GW of inference to help bring personal superintelligence to everyone. We would love all the feedback to keep hill-climbing. Please send it our way. We will keep hill-climbing on efficiency, user
3 relayeurs à 3 · il y a 8 h Voir sur X
Qui l'a repris
Grace Li” AI at Meta Arena.ai
Ce qu'ils en ont dit
Grace Li @grx_xce a cité · 09/11 20:43
Fast and mighty! Congratulations on the speed - both model and team.
7
Vals AI @ValsAI · 09/11 20:45
Meta's Ember is the cheapest model on the Vals Index at just $0.15 per test outperforming both Haiku 5.5 and GPT-6 Luna. It also finishes tests in about 15 minutes, making it the fastest model in the top 12.
3 relayeurs à 3 · il y a 8 h Voir sur X
Qui l'a repris
Alexandr Wang” AI at Meta Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)”
Ce qu'ils en ont dit
Alexandr Wang @alexandr_wang a cité · 09/11 20:45
"[Muse Ember🔥] also finishes tests in about 15 minutes, making it the fastest model in the top 12." super duper fast model ⚡️
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) @teortaxesTex a cité · 09/11 19:31
token-efficient top-tier Flash model, wow
8
S
Sam Pasupalak @spisallyouneed · 09/11 18:23
Introducing Steve: the World’s first AI CEO for fully Autonomous companies. Powered by World Models, Recursive Self Learning, a Policy Engine and a team of specialist AI agents, it runs your business end-to-end 🔥. Proof? Steve currently runs 3 businesses we acquired and they
3 relayeurs à 3 · il y a 10 h Voir sur X
Qui l'a repris
François Chollet” AK elvis”
Ce qu'ils en ont dit
François Chollet @fchollet a cité · 09/11 18:23
Sam and his team are working on a super exciting idea: autonomous businesses. Acquiring distressed or stagnant SaaS apps and handing the keys to an agent harness is the ultimate real-world eval. Just P&L as the objective function and the market as the feedback signal. Congrats
elvis @omarsar0 a cité · 09/11 17:38
Self-improving agents and world models are two ideas I am watching most closely. Skyfall AI put both into Steve, a harness built to run a business end to end. Steve records what went wrong after every loop and simulates the outcome of a decision before it acts. Skyfall has
9
Florian Brand @xeophon · 09/11 17:27
jammed with @HamelHusain a bit and since then, i cannot unsee of the state of evals as being similar to the military-industrial complex
3 relayeurs à 4 · il y a 11 h Voir sur X
Qui l'a repris
Sarah Catanzaro” elvis” Hrishi
Ce qu'ils en ont dit
Sarah Catanzaro @sarahcat21 a cité · 09/11 17:27
The solution seems pretty simple. We need better alignment between frontier labs and data vendors to track the core capabilities that matter so that agent/app builders can select models to test. Then we need to make it a helluva lot easier for teams to build internal evals that
elvis @omarsar0 a cité · 09/11 15:12
Own your evals, folks! I have nothing against public benchmarks; they are still important for research and public discussions. However, the only way to truly trust which models to use in prod is to build your own evals. And it's not just about measuring reliability and model