Bennett Bui ./

[Bennett Bui ./ @BennettBui1]

@0xEricYang @VinoisAsian your jobs

[Hexx ./ @HexxRL]

open source models demonstrating “your margin is my opportunity” in 4k

you are gambling with every prompt anyways for output, might as well go cheap with more tries.

[Hexx ./ @HexxRL]

🧩@Gradient_HQ Puzzle Mastermind!

Play from now til June 6th 1AM EDT!

🏆 Top 10 Participants will receive Quiz Mastermind on Discord!

Rules:
- username must match DC username
- play as many times & anytime you want during event duration

./ puzzle here: https://t.co/TWMxUb6FTw

[Hexx ./ @HexxRL]

every tech cycle delivers 10x more compute

the top blasting is not stopping just yet

if there’s a shortage there’s always latent compute somewhere

[Hexx ./ @HexxRL]

locally operated AI agents will be the closest thing to taste like ubi if it’s not already in being pursued right now.

saving rates in america has evaporated once again, need additional hands and value add for income

[Hexx ./ @HexxRL]

“AI will replace all human jobs, everyone will be useless”

meanwhile many of the fear mongering labs selling this narrative are workforcemaxxing with their headcount.

there’s some truth to having repetitive work eliminated but alot of it is probably greatly exaggerated.

[Hexx ./ @HexxRL]

M3 autonomy in PostTrainBench just dusted everyone else.

If M3 didn’t beat in raw outperformance it beat in cost per intelligence. It’s number 3 only behind Opus 4.7 and GPT 5.5, everyone after is not close.

1M context too, can’t wait to see where autonomy is at eoy 😆

[Hexx ./ @HexxRL]

Uncommonroute trained router matches Claude Opus 4.6 in SWE-bench Verified evaluation.

In TwinRouterBench you test realism agentic trajectories and the results are staggering:

Uncommonroute Trained vs Claude Opus 4.6

75/100 vs 74/100 (matched in resolution)
$25.66 vs $54.73 (53% cost saving with Uncommonroute trained)

Models cost more now with advanced reasoning and agentic tasks, time to save to get same quality at a better price.

[Bennett Bui ./ @BennettBui1]

We never end.

[rw ./ @gradientintern]

Hello everyone. X is removing it’s X communities feature on May 30th

I started building this community in January 2025 and it’s been a wonderful experience to meet nearly 5,000 of you inside of our Open Intelligence Lounge

@Gradient_HQ will continue to update it’s research efforts on main page and on Discord with other activities as well: https://t.co/YpRU7OHeux

Let’s stay in touch and there will be more to come!

[Hexx ./ @HexxRL]

i was the fourth member of lounge (**wink wink** 😏)

anyways join @Gradient_HQ discord as well here:
https://t.co/62qQYPjSHP

see you guys there!

[Bennett Bui ./ @BennettBui1]

#AI

[Hexx ./ @HexxRL]

SBF was out here aping OpenAI back in 2018 with Alameda writing a 500k check

@SBF_FTX is literally a generational VC player, got in Anthropic and OpenAI very early days.

His legacy would be so different if funding sources were from proper avenues

[Commonstack @commonstack_ai]

Run Claude Code with Commonstack in 4 steps:

- generate an API key
- set 4 environment variables
- run claude
- /status to verify

Set it up now in 5 minutes with @alex_mirran.

[Bennett Bui ./ @BennettBui1]

@HexxRL @Gradient_HQ ./ Grad roll

[Hexx ./ @HexxRL]

gradient bringing open infra into the agentic era.

coordination, privacy and accessibility have been strong research efforts from the lab.

happy to see gradient with other leading companies and labs powering agent infrastructure.

./ time to roll @Gradient_HQ

[Bennett Bui ./ @BennettBui1]

@gradientintern @Gradient_HQ ./

[rw ./ @gradientintern]

./ @Gradient_HQ

[Hexx ./ @HexxRL]

xAI has a lot of underutilized capacity that could be used as a white labeling service for the neoclouds who haven’t finished building their infra, fulfill those signed commitments to boost rev

also distributed training is music to my ears cc @Gradient_HQ

[rw ./ @gradientintern]

open source models so good now paid fudding in action.

once you have your own machines you don’t need to fully depend your entire system on a few providers blasting you with constant usage limits

[rw ./ @gradientintern]

H100s rental prices have risen around 40% in the past 2 quarters

Hovered from $1.70-$1.80 to now $2.35-$2.50. H100s are nearly 4 years old and finding massive demand.

Post training, agents and foundational models can’t get enough compute.