Hammer
877 posts
Hammer
@_hammersi
Consolidated to a mere vibe coder - AI, experimenting and enjoying the new era.
vibe world Katılım Nisan 2022
936 Takip Edilen1.4K Takipçiler
@jun_song We are at the iPhone stage
Every new version is incremental upgrade.
Every so often a new breakthrough like the m series processors will arrive and we will have some jump.
We can now optimize inference significantly which it self will bring massive improvement
English
To anyone expecting Opus-5 to make a massive leap over Fable:
Sorry to break it to you, but we’ve been hitting a wall in AI progress for months now.
There are no revolutionary breakthroughs. Everything from data curation and weight compression to compute efficiency has stalled out.
We like to think we are accelerating, but looking at it objectively, the progress from Opus-4.5 to 4.8 is tiny—it’s mostly just harness improvements.
In the end, every company chose to force-feed their weight sizes simply as a defensive move for their massive funding.
Starting from Fable, to Sol, Grok, and literally all the Chinese AIs—they all just bloated their weights.
This brought massive inefficiency to data center racks and only worsened the already scarce compute.
And that is precisely what gave latecomers like Chinese labs and xAI the time to catch up.
Due to this bottleneck, the first-mover advantage that frontier labs used to enjoy is rapidly shrinking.
Now, the race has simply reset back to the same starting line: whoever brings the next real innovation wins.
English
I don't fully understand why programmers seem to be the ~only group of people not angry that their jobs are being automated.
English
US AI model creates security issues.
Chinese AI model fixes them.
The same US model that created the bugs, refuses to fix them because of "cyber guardrails".
Think about it.
calle@callebtc
I have a report full of security issues of a software I'm working on. Codex won't fix them because of Cyber guardrails Fable won't fix them because of Cyber guardrails Kimi K3 fixed them all. No restrictions, just gets the job done. This will end badly for OpenAI & Anthropic.
English
The dissonance between the good guys and the bad guys..
One halting rev to serve existing users
The other rerouting you to dum models without you knowing and make you pay 100% of the price
Kimi.ai@Kimi_Moonshot
Kimi K3 has received far more love than we expected, and our GPUs are feeling it. Over the past 48 hours, demand has pushed close to the limits of our current capacity. To protect the experience of existing subscribers, we're temporarily pausing new subscriptions and prioritizing compute for current members. Existing subscribed users are not affected. We're adding capacity as fast as we can and will reopen new subscription spots in batches. Going forward, we'll also split membership into two more focused plans: Kimi Membership for Kimi Web, App, and Work; and Kimi Code Membership for coding workflows. This will help us match compute more precisely and keep the experience stable. Thank you for your patience and understanding!
English
I'm surprised this wasn't common knowledge
Sonnet is 1T parameters
Opus is 5T parameters
Fable/Mythos is 10T parameters
English
Kimi K3 has received far more love than we expected, and our GPUs are feeling it.
Over the past 48 hours, demand has pushed close to the limits of our current capacity. To protect the experience of existing subscribers, we're temporarily pausing new subscriptions and prioritizing compute for current members. Existing subscribed users are not affected.
We're adding capacity as fast as we can and will reopen new subscription spots in batches.
Going forward, we'll also split membership into two more focused plans: Kimi Membership for Kimi Web, App, and Work; and Kimi Code Membership for coding workflows. This will help us match compute more precisely and keep the experience stable.
Thank you for your patience and understanding!
English
Ironically, all security issues I need to fix were created by GPT 5.6 itself.
The same model that created the bugs, refuses to fix them because of cYbEr gUaRdRailS.
This is a complete clown show.
calle@callebtc
I have a report full of security issues of a software I'm working on. Codex won't fix them because of Cyber guardrails Fable won't fix them because of Cyber guardrails Kimi K3 fixed them all. No restrictions, just gets the job done. This will end badly for OpenAI & Anthropic.
English
@yacineMTB Let’s hope as a collective at least we will be able to adapt
English
@tengyanAI I Really hope that their poor and stupid oblivion marketing strategy will blow right on their faces
English
so let me get this straight
anthropic terrified us all to think they had a superhuman model that would wreck havoc on the world (bio, cyber) and imposed heavy limits on usage
turns out all the big labs have something similar too, and at 1/2-1/4 of the cost
so now let’s open it up and party?
Claude@claudeai
Beginning July 20, Claude Fable 5 will be included in all Max and Team Premium plans, at 50% of limits. Pro and Team Standard users will continue to have access to Fable via usage credits, and will receive a one-time $100 credit. Demand for Fable has been challenging to predict, which is why we rolled it out to subscription plans in stages, extending access several times as we secured additional capacity.
English