the most annoying part about claude, fable, sonnet, any of those models - is that's it's whining and refuses to work. "it'll take me 1 hour of automation to do that" - ok? i am paying for it, do it. getting tired of this.
nadir retweeted
Releasing the model weights and technical report of Kimi K3.
Kimi K3 is our most capable model: a 2.8T MoE model with native visual understanding and a 1M-token context window.
New model architecture: 2.5x the intelligence per unit of compute, not just more params.
Alongside Kimi K3, we're opening up more of the stack behind it — high-performance attention kernels, MoE communication library, and infrastructure for running agent environments at scale.
Model weights: huggingface.co/moonshotai/Ki…
Tech report: github.com/MoonshotAI/Kimi-K…
Tech blog: kimi.com/blog/kimi-k3
nadir retweeted
Kimi K3 is running 100% FASTER than it was this morning.
Why? Moonshot hit capacity and chose to stop selling NEW subscriptions instead of throttling existing ones. Every plan: sold out. On purpose.
When Anthropic hit the same wall in April, they cut existing users' usage 50% during peak hours.
Moonshot cut their revenue. Anthropic cut your usage.
That says everything.
My Kimi K3 subscription is not going anywhere.
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!
Been using Kimi K3 for the past day, it makes claude opus look like a donkey. Kimi is fast, reliable, smart.
Kudos to the Moonshot AI Lab.
I almost believed that claude will be able to pull off Fable's release with minimal issues.. i was wrong
Booted up my computer, used Fable 5 to compact an existing session => 10% of my session usage
I have 4 more sessions that need a compact, i guess i won't be doing it with fable
skills are not needed when you work alone on a project. Use skills when you have a repeated action to are performing on a cadence and you need repeatability, or when working in a group and you want to teach the other's agents to start using specific tools
love the new claude code feature where your text becomes cryptic and cannot be cleared even if the session is stopped and resumed.
Permanent memory is the worst feature across all agents.
“Claude is not as good as it was”
Memory is a system prompt that the agents create for themselves. You have no idea what’s written there, how correct it is and how it affects agents performance. Usually it’s for the worse
I should be able to take a screenshot without my device actively working against me.
nadir retweeted
This solo indie dev spent 10 years making an RPG game that looks 3D… but is actually 2D.
- Camera shifts from orthographic to perspective views
- Depth is simulated using parallax & warped sprites
- Hand-painted semi-open world
Would you play this? It’s called Bridgebourn.
nadir retweeted
AI can now generate pixel-perfect animations with frame interpolation
> Generate any pixel art sprite
> Create an end-frame variation
> Guide the animation with a prompt
The animation model will look at the prompt, and generate any missing frames
What can it be used for?
> Frame start-end guidance
> Regenerating part of an animation
> Chaining multiple animation together
nadir retweeted
Yes, it's completely dead. All cybersecurity people I know have quit their jobs and transitioned into being anime vloggers
nadir retweeted
Appears that a big chunk of Claude Code's source code has been exposed on npm via a .map file accidentally uploaded to the public registry.
~512K lines of code
~1,900 files
HugOps to the Anthropic team, this is brutal
github.com/instructkr/claude…
we are living in a revolution and few realize it yet
