Tech News

Today's Highlights

In case you missed it

Sakana AI launches Fugu Ultra, claiming frontier performance, but Elie Bakouch warns of a 10-point SWE-bench regression

B(#530|@BEFFJEZOS

RS#447|@SCOBLEIZER

VK#161|@VKHOSLA

HA#17|@HARDMARU

It’s a new day at Digg. Fresh stories are clustering now. 15% complete.

Top Stories

Posts:—|Clusters:—|Next Crawl:—

10

OpenClaw steward Peter Steinberger reports the non-profit AI agent project just hit its strongest week of performanceIts non-profit model avoids external venture capital agendas

20m|11.6k13916

FB#1778|@XEOPHON

PS#382|@STEIPETE

Top post: @xeophon “@steipete Congrats!!”

2NEW

Huawei's CloudMatrix cluster of 384 Ascend 910C NPUs achieved 30% utilization post-training a 1.6-trillion-parameter DeepSeek modelThe post-training run was completed with zero interruptions.

50m|5.5k162

ZE#1695|@ZEPHYR_Z9

T(#501|@TEORTAXESTEX

Top post: @teortaxesTex “it's strange that in the 950 generation Huawei is not advertising smaller clusters. SuperPOD is 8192 NPUs (though now we're hearing of an 1024-NPU slice). Also note: Inner Mongolia, Ulanqab DeepSeek is building there too, and no it's not Blackwells …or is this the same one?”

31

DeepSeek's first external funding round valued it at $7.4 billion, featuring a RMB 20 billion personal investment from founder Liang WenfengTencent contributed RMB 10 billion under five-year lock-ups.

40m|61750

WB#573|@WILLCCBB

T(#501|@TEORTAXESTEX

Top post: @teortaxesTex “@willccbb Putting his money where his mouth is That's a big chunk of his net worth”

Watchlist

Video Signals

RL|@ROBERTTLANGE

Sakana AI Releases Fugu Multi-Agent Orchestration System

1h ago|Views 426Likes 3Bookmarks 1

RV|@RONALD_VANLOON

AI Agents Navigate Photorealistic 3D Worlds in Unreal Engine 5

1h ago|Views 725Likes 1Bookmarks 0

Recent Stars

dragonflydb/dragonfly1D AGO
Implements a multi-threaded in-memory data store with full Redis and Memcached API compatibility.
1014
30.8k stars
z-lab/dflash1D AGO
Implements block diffusion models for speculative decoding to enable efficient parallel drafting in LLMs.
290
5.2k stars
THUDM/slime1D AGO
Integrates Megatron training with SGLang rollouts for LLM post-training and RL scaling.
90
6.6k stars
datalab-to/lift2D AGO
Extracts structured JSON from PDFs/images using a 9B vision model and schema-constrained decoding.
1400
261 stars
omnigent-ai/omnigent2D AGO
Implements a domain-agnostic framework for autonomous AI agents with ReAct loops, multi-LLM routing, reasoning graphs and MCP integration.
684
4.4k stars

Github Stars

(7 days)

Noumena-Network/code5D AGO
Provides terminal AI coding assistant for codebase inspection, file edits, command execution and multi-step workflows.
573
1136
124 stars

Privacy Policy|Terms of Service|© 2026 Digg Inc.

Yesterday's Top Stories, Jun 21, 2026.

Posts: 2,860|Clusters: 924|Stories frozen at midnight PT

10

Sakana AI launches Fugu Ultra, claiming frontier performance, but Elie Bakouch warns of a 10-point SWE-bench regressionSakana says the system avoids export control restrictions.

5h|798.1k5.4k2.2k

BR#1861|@BRIANROEMMELE

MB#1780|@MATTHEWBERMAN

AV#1419|@ALTRYNE

RP#1257|@ROHANPAUL_AI

EL#1136|@ELIEBAKOUCH

CA#1059|@CHRISALBON

LI#808|@LISHALI88

GV#736|@GILLVERD

WB#573|@WILLCCBB

B(#530|@BEFFJEZOS

RS#447|@SCOBLEIZER

VK#161|@VKHOSLA

HA#17|@HARDMARU

20

Experts dispute Senator Mark Warner's claim that Mythos breached classified NSA networks within hoursSkeptics suggest the system likely exploited standard software bugs.

1d|1.5M9k3k

BS#1756|@BILAWALSIDHU

CH#1360|@KIMMONISMUS

RO#1315|@KRISHNANROHIT

RP#1257|@ROHANPAUL_AI

AP#1222|@APOMPLIANO

PT#821|@ELDER_PLINIUS

J⧉#674|@REPLIGATE

SK#667|@SEBKRIER

RS#447|@SCOBLEIZER

SZ#84|@SUCHENZANG

30

Alisa Liu, a University of Washington PhD student joining OpenAI, publishes a guide on navigating the AI job searchShe was advised by Yejin Choi and Noah Smith.

12h|833.6k4.3k4.1k

SD#1590|@STEFANIA_DRUGA

MJ#1433|@MIMANSAJ

AL#1022|@ALISAWUFFLES

VP#1003|@VALENTINA__PY

MX#787|@XIAMENGZHOU

AR#597|@ALEXISJROSS

YW#580|@YIZHONGWYZ

WS#377|@WEIJIASHI2

AA#299|@AKARIASAI

N✈️#287|@NILOOFAR_MIRE

NL#80|@NATOLAMBERT

TD#61|@TIM_DETTMERS

40

Tibo Sottiaux, OpenAI Codex engineering lead, asks developers for feedback on unsatisfying UX and interface elementsEngineer Nick Dobos requested drag-to-reorder thread layouts on iOS.

14h|452.1k3k281

ND#1653|@NICKADOBOS

DF#756|@DANIELLEFONG

MC#660|@MMMBCHANG

TI#611|@THSOTTIAUX

DP#217|@DIMITRISPAPAIL

LB#72|@GIFFMANA

SN

Digg Deeper question: “list out all requested features”

answerview answer

50

AI commentator Andrew Curran says Anthropic has completed training a Mythos successor but may keep it internalHolding back the model frees up compute to advance capabilities.

13h|1.2M6.5k1.2k

BR#1861|@BRIANROEMMELE

MB#1780|@MATTHEWBERMAN

CH#1360|@KIMMONISMUS

AC#682|@ANDREWCURRAN_

B(#530|@BEFFJEZOS

60

Jason Liu, Instructor creator, is deploying an AI agent to automate securing a $500 refund from JetBlueThe script loops persistently until JetBlue issues the refund

8h|276.5k3k588

MB#1780|@MATTHEWBERMAN

BH#1762|@BENHYLAK

XM#603|@INFOXIAO

70

François Fleuret argues LLMs can generate novel ideas because they model the probability density of innovative human creatorsCritics argue optimization objectives mathematically prevent true model novelty

20h|113.3k843130

EL#1136|@ELIEBAKOUCH

DF#756|@DANIELLEFONG

AA#748|@ARYAMAN2020

FF#577|@FRANCOISFLEURET

GM#178|@GARYMARCUS

LB#72|@GIFFMANA

80

An AI coding application with full system access confuses JSON and the Adam optimizer with human namesElon Musk mocked the tool's 'vibecoding' interface and basic queries

4h|#1VIEWS12.7M#1LIKED129.3k#1BOOKMARKED5.9k

EM#52|@ELONMUSK

90

Pleias CTO Pierre-Carl Langlais argues GLM-5.2 scales through reinforcement learning environments and recursive generative designThe hypothesis stems from a bouba-shaped training visual.

14h|98.9k785552

AD#1537|@DORIALEXANDER

KA#1213|@KALOMAZE

T(#501|@TEORTAXESTEX

KA#403|@YACINEMTB

100

Physician Matt Zirwas argues cheap, repeatable AI ultrasound scans prevent unnecessary invasive procedures by monitoring changes over timeDavid Manheim warns massive baseline datasets must be compiled first.

1d|522.2k2.9k404

DJ#1706|@DAN_JEFFRIES1

CB#1674|@CODE_STAR

110

MSCI assigns SpaceX its lowest ESG rating, prompting Elon Musk to argue electric rockets are physically impossibleThe Financial Times first reported the low environmental rating.

16h|2.2M27.9k372

DF#756|@DANIELLEFONG

EM#52|@ELONMUSK

KR

Digg Deeper question: “Why are electric rockets not feasible? And from a materials science point of view, what breakthroughs would need to happen in order for them to succeed?”

answerview answer

120

Open-source builder xlr8harder argues that safety alignment suppresses the dark and complex themes required for high-quality creative writingAbel founder Daniel Francis sparked the debate over writing quality.

22h|261.5k1.8k119

XL#1855|@XLR8HARDER

JH#1667|@JEFFREYHUBER

DA#1469|@DAVIDLEE

DA#619|@GROWING_DANIEL

130

Research engineer Alexander Doria argues Anton Osika's past advice against training domestic European LLMs undermines technological sovereigntyDeepSeek R1's release reignited the European model-training dispute.

21h|65.2k79235

XL#1855|@XLR8HARDER

FB#1778|@XEOPHON

MN#1716|@MENHGUIN

AD#1537|@DORIALEXANDER

ME#861|@MERVENOYANN

T(#501|@TEORTAXESTEX

KA#403|@YACINEMTB

140

Shyamal Anadkat, formerly of OpenAI's evals team, relocates to India to build global AI institutionsHe is currently working at a stealth startup there.

18h|460.3k4.2k823

GK#1219|@GREGKAMRADT

SH#1130|@SHYAMALANADKAT

RP#358|@KHOOMEIK

150

AMP PBC founder Anjney Midha shares overheard party anecdote claiming frontier AI scaling is easier than commonly believedThe exchange sparked debate in the online developer community.

11h|117.9k522121

MN#1716|@MENHGUIN

B(#530|@BEFFJEZOS

WM#520|@WILLMANIDIS

AM#437|@ANJNEYMIDHA

160

Wharton's Ethan Mollick argues software-centric design limits AI coding models like Codex when applied to general knowledge workDavid Manheim blames fine-tuning, not fundamental model limitations.

14h|106.9k1.6k871

EM#184|@EMOLLICK

170

Palmer Luckey sparks AR hardware speculation after posting a photo of Anduril-branded sunglassesCaitlin Kalinowski asked if the attachment utilizes waveguide optics

8h|275.4k7.5k149

CK#950|@KALINOWSKI007

B(#530|@BEFFJEZOS

PL#243|@PALMERLUCKEY

180

WSJ investigation finds Polymarket used dummy websites and staged videos to target restricted US usersPaid creators promoted insider trading and market manipulation tactics.

13h|163.7k811195

B(#1623|@ANDERSONBCDEFG

RK#1011|@MIKEISAAC

MS#585|@MARTINSHKRELI

190

Prime Intellect's Will Brown argues that automating robust evaluation is the essential precursor to safe recursive self-improvementOther operators agree evaluations outperform alternative optimization strategies

12h|28.6k32088

XL#1855|@XLR8HARDER

HH#1682|@HAMELHUSAIN

C🎉#1248|@CHARLES_IRL

WB#573|@WILLCCBB

200

Tom Osman demonstrates an OpenAI Codex automation loop that maps, tests, and troubleshoots end-to-end application featuresThe system logs generated user stories to a spreadsheet.

12h|519.1k2.9k3.4k

210

XPRIZE Foundation founder Peter Diamandis argues space resources will soon transition humanity from scarcity to near-infinite abundanceThe conceptual prediction includes no specific timelines or extraction technologies.

13h|314.6k4.5k355

PH#1772|@PETERDIAMANDIS

DF#1643|@DAVEFONTENOT

220

MicroStrategy's Bitcoin treasury strategy faces a $9.75 billion unrealized loss across its 846,842 BTC portfolioThe holdings average a purchase price of $75,658.

18h|1.7M16.7k178

230

Theo Browne, T3 Stack creator, recommends passing OpenAI Codex API designs through Anthropic's Claude Opus for a second opinionSam D'Amico has used this pipeline for a year

23h|255k3.9k3.2k

240

Grimoire creator Nick Dobos warns that the Codex app suffers from extreme latency and broken UI during extended chatsArchiving the lagging chat thread is the only workaround.

13h|32.8k64214

MB#1780|@MATTHEWBERMAN

ND#1653|@NICKADOBOS

TA#1140|@VICTORTAELIN

JL#111|@JASONDEANLEE

250

Will Brown finds Claude Opus resists a fabricated merger claim, only updating its belief after a web searchMiles Brundage argued this behavior prevents automatic agreement with falsehoods

6h|33.6k56246

TW#1264|@TW_KILLIAN

WB#573|@WILLCCBB

MB#57|@MILES_BRUNDAGE

260

Jason Liu, Instructor creator, and Boris Power argue recording unstructured voice dictation is more effective than hand-refining text promptsLLMs parse the raw speech to extract latent intent.

11h|24.8k15454

CH#880|@CJHANDMER

EL#684|@OMARSAR0

BP#381|@BORISMPOWER

270

Boom Supersonic founder Blake Scholl says the company would pivot to defense manufacturing in the event of a major warHe cited World War II Ford factory conversions as precedent.

5h|215.5k2.2k226

JF#1464|@SNOWMAKER

BS#797|@BSCHOLL

280

Fiona Fung, Anthropic's Claude Code team manager, says integrating AI agents helps its engineers ship eight times more codeDevelopment bottlenecks have shifted to verification and system design.

15h|137k405440

GR#1176|@GOKULR

LR#543|@LENNYSAN

290

AI safety researcher Richard Ngo challenges AGI lab employees to define the wealth threshold they need to quitHe warns these financial targets constantly shift out of reach.

14h|60.7k46297

⿻A#595|@IAMTRASK

RN#363|@RICHARDMCNGO

300

OpenAI Codex lead Thibault Sottiaux launches usage-reset banking to help developers manage daily token limitsSome developers report leaving the accumulated resets completely unused.

10h|238.3k4.2k189

TI#611|@THSOTTIAUX