.@browsercompany's arc browser has been crashing consistently 2-3x per day on me
anyone else getting this issue
looking for 3-6 b300 nodes on a 3-6 month term, live asap for pareto inference
- bare metal, us east
- amd epyc, 3tb+ ram per node
- same dc, shared ib/roce fabric w gpudirect rdma
- under $6/gpu-hr
dms open
dylan vu retweeted
Pareto now offers zero data retention: we don't keep your prompts or outputs after a request finishes.
paretoinference.com/zdr
if you're long gravel for ai, pivot to water
BREAKING: @RainmakerCorp has secured a deal to bring cloud seeding to data centers.
The unnamed data center provider will use Rainmaker to generate additional rainfall and replenish freshwater supplies, according to the @sfstandard.
The El Segundo startup uses autonomous drones, weather modeling software, radar, and satellites to seed clouds and measure the precipitation it generates. Rainmaker is currently running what it calls the largest cloud-seeding operation in modern American history across northern Utah and southern Idaho.
dylan vu retweeted
$3/week for GLM and DeepSeek API access
Pareto Pass is officially live 🎉
70 tok/s median on GLM-5.3
Join with a referral: $1 for your first week
Refer a friend: earn $3 subscription credit after their first $3 renewal
Get started: paretoinference.com
🤖 Made with AI
we're giving out free glm 5.3 inference to get experience as a provider. reply if you want access. limited seats
GLM-5.3 is now open-weight.
Our most capable model for agentic coding and cyber defense is now available to download, run, and customize.
Weights: huggingface.co/zai-org/GLM-5…
Tech blog: z.ai/blog/glm-5.3
whenever i find myself doing something on my computer that i don't interact with every day, like
>make discord server
>use litellm
>dns shit btwn vercel and namecheap
i'm using codex computer use
phenomenal experience (we must go faster)
.@claudeai @ClaudeDevs is so buggy and broken. it's painful to use. night and day diff versus @ChatGPT desktop app
i'm constantly experiencing:
-your messages weren't sent, try again later errors with my prompts disappearing from the input field (HUGE time loss to recreate the prompt)
-network errors that both disallow me from seeing the response loading and also from sending a new prompt (i can't see when chats clear up for new inputs)
-response timers resetting randomly
etc.
frontier models for indeterminant, idea maze exploration work
pareto frontier models for determinant, repeat workflow work
the guys over at @USDai_Official corroborate this on the hardware side
not online are they finetuning models, but are also purchasing gpus as demand matures for improved per token margin
