rita kozlov ๐ retweeted
one thing that made this click for me โ with decision models, you have to define categories and it returns probabilities that the input matches those categories.
but so much of the data we interact with is completely new to us. how can you define categories for something you've ic, you can extract values directly from the inputs, without any premeditated categories. the pipeline becomes clairic to extract values โ clef to make decisions with those values โ LLMs to take actions on those decisions. all of these models run on Workers AI.
but so much of the data we interact with is completely new to us. how can you define categories for something you've ic, you can extract values directly from the inputs, without any premeditated categories. the pipeline becomes clairic to extract values โ clef to make decisions with those values โ LLMs to take actions on those decisions. all of these models run on Workers AI.
a brand new model paradigm! introducing ๐๐๐๐๐๐๐: a system one model for extracting, computing, estimating, and predicting values from unstructured inputs.
blog.cloudflare.com/clairic-extracโฆ
23
65
898
40,457
rita kozlov ๐ retweeted
Today we introduced a novel system one model, Clairic, based on regression instead of classification (i.e. decision models like Clef). This enables estimation of continuous values without having to exhaustively list choices upfront. It also enables very efficient extraction of structured inputs. Open weights coming soon!
a brand new model paradigm! introducing ๐๐๐๐๐๐๐: a system one model for extracting, computing, estimating, and predicting values from unstructured inputs.
blog.cloudflare.com/clairic-extracโฆ
33
78
1,089
127,735
rita kozlov ๐ retweeted
There are some really interesting things you can do with this and our Basin analytics platform. I'm going to cook up a demo tonight - come find me at the Cloudflare booth tomorrow to check it out!
we shipped a new type of model: an extraction model!
pass it unstructured data, and get a list of the details and probability back. try it out and let us know what you think!
blog.cloudflare.com/clairic-extracโฆ
0
3
12
1,377
we shipped a new type of model: an extraction model!
pass it unstructured data, and get a list of the details and probability back. try it out and let us know what you think!
blog.cloudflare.com/clairic-extracโฆ
pass it unstructured data, and get a list of the details and probability back. try it out and let us know what you think!
blog.cloudflare.com/clairic-extracโฆ
23
36
484
28,648
rita kozlov ๐ retweeted
Workers Issues can detect and classify vulnerabilities scanning on your app
If you get these pls go turn on waf u silly goose ๐ฅฐcloudflare.com/products/waf/
If you get these pls go turn on waf u silly goose ๐ฅฐcloudflare.com/products/waf/
0
3
14
1,598
oops
So many people flying in for @Cloudflare Connect the SFO airport is literally unable to handle it
4
2
35
2,103
rita kozlov ๐ retweeted
Headed to SF for @Cloudflare Connect. Find me after my talk Tuesday at 1:50pm if you want to say hi. Thinking about slop forking something live on stage. What should it be?
10
2
77
4,082
love to see region: earth still going strong after all these years
the cloud region picker is so iconic, this was the best way to communicate the thing that was so unique about workers. you donโt have to pick a region, itโs just earth
the cloud region picker is so iconic, this was the best way to communicate the thing that was so unique about workers. you donโt have to pick a region, itโs just earth
Iโve always really liked the Region: Earth copy btw
Hello, SF
3
3
76
2,643
new @pipecat_ai phonellm model with cloudflare workers (in python) + webrtc
Did you know you can run a Pipecat agent on Python Workers? We didn't either until we made this:
github.com/cloudflare/reaโฆ
It uses WebRTC for transport thanks to Cloudflare Realtime SFU. And I also added some nice keyboard sounds for when the agent looks things up.
And yes, I what's one more Matt at Cloudflare?
3
1
13
1,395
rita kozlov ๐ retweeted
Amused that AWS just unilaterally cancelled Cloudflare's Reinvent booth. There are always signs when someone finally realizes they're losing. Guess we'll just attend and throw the most fun side events without paying Amazon! Reply if you want an invite.
437
151
3,861
537,856
so much aura ๐ฎโ๐จ
keep the good vibes going next week at cloudflare connect cloudflare.com/connect
keep the good vibes going next week at cloudflare connect cloudflare.com/connect
Well this is fun ๐งก
3
2
57
3,378
rita kozlov ๐ retweeted
For your weekend reading: Build with Pi Durable on Cloudflare, a walkthrough of how to create a multiplayer chat agent, including subagents, on Durable Objects.
earendil.com/posts/pi-durabโฆ
earendil.com/posts/pi-durabโฆ
41
132
1,610
112,333
rita kozlov ๐ retweeted
Very excited partner with the Pipecat team to get PhoneLLM Mini on Workers AI.
Congratulations @kwindla and team - love using Pipecat and this model is fantastic. And huge thank you to @michellechen and the Workers AI team for making everything come together.
Congratulations @kwindla and team - love using Pipecat and this model is fantastic. And huge thank you to @michellechen and the Workers AI team for making everything come together.
Fast is better than slow.
We've got two new PhoneLLM models optimized for conversational voice AI.
PhoneLLM Mini (30B) โ When you need the lowest possible latency (sub-100ms TTFAT), smallest model size, and lowest cost. Good multi-turn conversation behavior and tool-calling t can run >100 concurrent agents on a B200 and fits on an RTX 5090.
PhoneLLM Max (120B) โ A "microthinking" model for complex conversational tasks. Sub-200ms TTFAT. SOTA task completion performance at 1/10th the latency and cost of models with similar capabilities.
Both of these models are full-weights fine-tunes of @NVIDIAAI Nemotron models. Mini is based on Nemotron 3 Nano. Max is based on the brand new Nemotron 3.5 Super VL.
Both of these models are available as open weights on HuggingFace, and from four inference partners: AWS, Baseten, Cloudflare, and Modal.
2
6
22
2,045