dax

@thdxr • 136,666 subscribers

building @opencode at @anomalyco not the ceo

Shorts

434,362 views

114,853 views

662,971 views

655,190 views

254,889 views

55,863 views

286,757 views

647,573 views

434,481 views

188,453 views

80,952 views

67,008 views

79,136 views

36,050 views

44,698 views

42,478 views

30,491 views

45,054 views

35,773 views

30,523 views

Videos

sweetdream.ai

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Private Show

Join now for exclusive access

Free preview available • Premium content

409,071 views • 1 month ago

125,558 views • 28 days ago

229,034 views • 4 months ago

1,153,109 views • 2 years ago

192,760 views • 5 months ago

285,724 views • 8 months ago

152,293 views • 4 months ago

199,247 views • 8 months ago

209,967 views • 10 months ago

219,506 views • 1 year ago

122,609 views • 9 months ago

66,232 views • 5 months ago

123,911 views • 1 year ago

109,046 views • 1 year ago

82,647 views • 11 months ago

64,637 views • 10 months ago

91,362 views • 1 year ago

83,626 views • 1 year ago

46,120 views • 9 months ago

Live Cam

dax

Shorts

we've improved our traffic routing for kimi k2.5 (paid version) on zen so it should now be absurdly fast might hit rate limits here as we scale up but this feels like something different

my waymo is stuck going round and round the roundabout

it's getting to the point where i can just feel that an app is using RSCs when i use it - nearly 1s page navigations i'm sure it can be improved but yet to come across an RSC app in the wild that feels good about to say v0 but just checked idk if they're using them anymore

haha idiots they implemented the paywall on the frontend so i can just inspect ele—

watch opencode absolutely destroy claude code using the exact same model

ok my work is done

we are officially working on opentui - a library for building terminal UIs in typescript it is built on the work of kmdrfx - we are sponsoring him to solve this problem once and for all clip is not sped up and it's running entirely in the terminal - link to repo in reply

after a decade of trying i finally found a cable management approach that i'm satisfied with i'll drop some tips in the thread

today we're finally giving you a solution to websockets our Realtime component is now available - connect millions of clients (like you even have that many) over websocket and broadcast messages to all of them it's ridiculously fast - look

the levels of jank being shipped has gotten pretty bad before you reply with the excuses i know - nextjs can deliver good experiences - this is related to app router migration but idc - end results are bad industry should shut up about how to build apps till this is fixed

someone submitted a pr that added "beast prompt" to opencode for gpt models it makes 4.1 actually good at tool calling look how fast this thing goes, video not sped up

playing with some speed improvements - hopefully can ship to you next week it's crazy how fast these things can get

at this point i'm wondering if people are just baiting me

we're shipping a new spinner every week i guess

made a general select dialog with opentui + solidjs it's a decent code example to get a sense of what things look like link in reply

some of you seem to care about silly things like making sure the LLM doesn't rm -rf / or delete your database i don't really get it but in opencode 0.3.102 we added configurable permissions you can be strict or specify what granular situations should prompt for permissions

opencode's /undo and /redo feature was pretty hidden - a lot of people don't know about how it can revert edits or bash calls in the new opentui version everything flows through the ctrl+p command bar...but also you can scroll up and click on the message you want to revert to

vibe coding lets us play around with much weirder UI concepts

in opencode v0.3.53 we implemented message queuing so you can send a message while opencode is working and it'll update the LLM on the next opportunity - basically lets you resteer it in a different direction should feel pretty natural

imma give it a try

Videos

Watch Anya Live

tired of this misinformation so we made a video on the truth behind the anthropic vs opencode drama

preview of our minimal mode that doesn't run as a fullscreen TUI we're designing this carefully, it never rewrites your scrollback which means there's some tradeoffs but it'll be the only coding agent that doesn't have flickering or weird layout shifts

finally got around to setting up an always on opencode server so i can run sessions on any device from anywhere takes a few minutes - showed it off here

impressive performance shouldn't need daily essays telling you it's impressive it should just be impressive here's searching through 30,000 patients at 60fps from a real app customers pay 5 figures a month for RSCs won't help you do this

we finally shipped thinking in opencode - press ctrl+t to cycle through various thinking levels it's weird this came so late but it's a good example of how we ensure opencode stays extremely customizable while still working well out of the box quick video breakdown

this is why you think open source models aren't good on the left is the average provider, on the right is opencode zen left is slow, vomits out tool call parse failures, confuses itself and takes 5x longer to complete

in OpenCode v1.1.18 we implemented the detailed planning flow you find in cursor or claude code but not 100% sure about it yet so we put it behind a OPENCODE_EXPERIMENTAL_PLAN_MODE=1 feature flag quick video on how it works but if you have feedback please put it on this thread

today we're launching opencode zen zen provides the best LLMs for coding ALWAYS in the best possible configuration no more "did claude get dumber?" no more being routed to a crappy provider and we're giving it to you at cost and you can use it anywhere that enough for you?

holy shit groq + k2 is fast af - video not sped up stuff is changing real fast

here's yet another app built on the side that - feels better than what most full time teams ship - is real, not a tech demo - screen built mostly by someone with &lt; 1yr of exp - no PhD in any framework required outcome is all that matters - your grand theories do nothing for me

been using a stealth model all day in opencode and it's working real well now you can use it too - it's called Sonic, it's made for coding and it's free for you to use so the team behind it can test it update to latest opencode and it should pop up right there

all this scrollback talk people keep @ ing us like we can't do it this has been in opentui for five months and it no it didn't take several months of work and a billion opus tokens pick better heroes and try to scare us for once

i rotated my credit card and my youtube sub broke and i started seeing ads holy shit is this what normal people are seeing? this is straight out of idiocracy

here's a practical case of how i use LLMs to code day to day - different than the "shit out a whole app" goal everyone is chasing

opencode now integrates with vscode, cursor, windsurf, etc it'll be aware of what file you're in and you can easily send inject selected lines into the prompt start opencode inside your editor and it'll install the extension

finally made a video on how and why i do all my development on a remote server i absolutely love it full video in reply

i've been trying to figure out how to articulate this over text but turns out a video is easier

we polished up our agents feature and it's real nice now - built in build/plan agents - tab to switch primary agent - @ mention to create subagent - automatic agent triggering simple config to make any combination of models, prompts, tools, whatever works for your workflow

here's yet another app built on the side that - feels better than what most full time teams ship - is real, not a tech demo - screen built mostly by someone with < 1yr of exp - no PhD in any framework required outcome is all that matters - your grand theories do nothing for me