Pinned Tweet
we got the full Kimi K3, 2.8T params, running on 80x RTX 5090s.
20 tok/s single stream, day one, untuned. Last week we took GLM-5.2 from 30 to 110 tok/s on this same fleet. This number will climb.
A first for open weights: frontier intelligence served with zero HBM, the scarcest silicon in AI. Just GDDR7 gaming cards, plain ethernet, and the official MXFP4 weights, nothing requantized.
The most powerful open model on Earth, on the most abundant GPUs on Earth. Any lab, startup, or university can now own it, probe it, fine-tune it, run agents on it.
@Kimi_Moonshot
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
Ning retweeted
DeepSeek-V4.1-Flash-1017 is live on engy.ai.
🚀 DeepSeek-V4.1-Flash-1017 is here!
Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API.
🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model.
1/4
Ning retweeted
Yesterday engy.ai processed 50B+ tokens in a single day, and we’re just getting started.
check our stats here: engy.ai/usage
congrats @modal
Today we're announcing that Modal has crossed $1B in annualized run rate, as well as raised a $750M Series D at a $15.75B valuation, led by @Accel.
Since our Series C, usage has tripled and we’ve grown more than 15x year-over-year.
We are deeply grateful to the developers and companies building on Modal. There's still so much more to build, and we can't think of a better time to be building software.
Ning retweeted
Get your ENGY.
Use my code: engy.ai/r/markjeffrey
30% cheaper than OpenRouter (or anywhere) for GLM 5.3 flash.
Cheapest inference on planet earth - with no compromise on speed or quality.
GLM-5.3 is live on engy.ai at ~100 tps.
$0.98 in / $3.08 out / $0.18 cache per 1M, 30% below OpenRouter and the official API ($1.40 / $4.40 / $0.26).
Ning retweeted
Engy Referrals is live: invite friends to Engy and earn 10% of their API spend
👉 Get your referral link: engy.ai/dashboard/referral
Start inviting and earning today!
Ning retweeted
.@Pareton_ai miners achieved up to 54% faster AI inference on @engyai’s setup, using the same GPUs.
On the latest Novelty Search, Pareton explains how Bittensor’s Subnet 10 improves the software that runs AI models, with each winning optimization becoming the next baseline to beat.
The team also got a miner-developed optimization merged into @vllm_project, one of the leading open-source engines for AI inference.
Hosted by @const_reborn
Full episode in the first comment
Ning retweeted
Proud to be a day-1 provider on @Architect_Fi. Inference need real markets and we're bringing capacity to help build them. Congrats @BrettHarrison and team.
Congrats
We’re excited to launch Liquid Inference with some of the fastest-growing token producers in the ecosystem:
• Engy @engyai
• Pearl Research Labs @prlnet
• AkashML @akashnetAI
• Hyperbolic @hyperbolic_labs
• Boundless @BoundlessHQ
• Zro @zroai_
Ning retweeted
Proud on @engyai growing 100% MoM in revenue while not relying on OR or other distributor channels. We are processing 4-5b Kimi k3 tokens daily and scaling upwards. If you're still using OR, take some time to test it out.
Better pricing and better customer service
Ning retweeted
Introducing Beam: a highly efficient agentic open model with 501B total parameters and 23B active.
- Frontier reasoning efficiency
- Advances the Western open frontier on coding & agentic tasks
- Trained end-to-end from scratch
Full weights release this month.
Learn more about Beam: reflection.ai/beam
I truly believe the most important and powerful AGI system in human history will be run and organized in a decentralized way, and the protocol is most likely Bittensor.
As a subnet builder, I see endless possibilities in what this protocol enables. Bittensor has evolved a lot through real practice, and no other protocol has come this far.
Bitcoin decentralized money. Bittensor is decentralizing intelligence
$TAO 2030 is our case for Bittensor becoming a $1tn+ ecosystem by 2030
Ning retweeted
🚨 Kimi K4 Leak: Beat fable 5.1
>Moonshot AI is reportedly already planning its next monster model
>Kimi K4 may beat Fable 5.1 and GPT-6 Astra
>Moonshot is reportedly seeking more Nvidia Blackwell GPUs to train it
>K3 already packs 2.8T parameters, the largest open-weight model released to date
>K4 could push Moonshot's open-weight strategy even further
>Worth noting: K3 itself is already tangled in controversy — the White House accused Moonshot of accessing Blackwell chips via Thailand servers to dodge export controls, and of distilling Anthropic's Fable model into K3's training
No confirmed parameter count or release date ye
Ning retweeted
Engy is getting faster.
Pareton now tunes how @engyai runs its models. Quicker replies. Same API on your side.
Engy is one of the fastest-growing inference networks on @bittensor and beyond. A big customer with serious traffic. That is the workload we were built for.
Ning retweeted
One model per agent is a ceiling. Fusion lifts it: mix open models behind one endpoint, cheap for easy steps, strong when it gets hard, helpers that advise the main one. Tested: it beats the best single model.
SN11 Season 2 opens Sep 20, 00:00 UTC. trajrl.com
Ning retweeted
The TAM of these usecases will be trillions, looking forward to contribute with @engyai
we will need compute (@lium_io, @TargonCompute)
we will need inference (@engyai, @chutes_ai)
we will need data (@Hippius_cloud)
we will need knowledge (@teutonic)
If we are going to have free intelligence (@affine_io)
Ning retweeted
we will need compute (@lium_io, @TargonCompute)
we will need inference (@engyai, @chutes_ai)
we will need data (@Hippius_cloud)
we will need knowledge (@teutonic)
If we are going to have free intelligence (@affine_io)
congrats @fish_datura @lium_io . you've proven the revenue flywheel works.
The masculine urge to buyback + burn $1,000,000.00 worth of my token with our product revenue.
taomarketcap.com/extrinsics/…
