@brendonovich @opencode @metruzanca this forces me to always try to remember to switch to plan mode befor opening a new tab just so the new tab has the right model
English
Tama
22 posts
@brendonovich @opencode @metruzanca Thanks it has now been working well, An annoying thing ive noticed and always have to think about is lets say i switch to build mode and it switches models when i open another tab and it opens in plan mode itll use the last selected model( build model) instead of plan model
English
@prathamdby My defaults are luna max for planning and luna high for execution
English
New meta: GPT-5.6 Luna for planning and DeepSeek V4 Flash for execution.
prath@prathamdby
GPT-5.6 Luna available in OpenCode Go. 🤯
English
@MiaAI_lab How long would it take you to use up $4000 through the api... and there'll be cheaper and bigger models before $1000 is used up
English
@duummbass @Json589734 @kylebrussell cheaper not better luna knows a lot more stuff and is smarter, i dont see myself using DSV4F instead of luna max
English
Chinese models, dead
Sonnet, dead
Composer, dead
Gemini Flash, dead
Cognition@cognition
We've updated FrontierCode 1.1 to reflect new discounts for GPT-5.6 Terra and GPT-5.6 Luna. With these new costs, the GPT-5.6 series sits on the pareto curve of price/performance efficiency.
English
@duummbass @Json589734 @kylebrussell Deepseek uses a lot more tokens so luna ends up cheaper, im trying them out right now and luna is a lot smarter and checks its work just 2 sessions in and my project has already improved
English
English
English
English
@Json589734 @kylebrussell luna max still ahead and only slightly more expensive
English
@getsome_air Yes similar to this but for commit message, currently generate commit message sends a prompt to the current chat and uses selected model for the commit message, this isnt ideal cause selected model could be an expensive one.
English
@tama_tech8 Could you help us understand #2 and #3 a bit better?
For commit messages, would you like to select a different model and a custom prompt to generate them?
For models, being able to save favorites makes total sense!
English
JetBrains Air@getsome_air
English
@tama_tech8 @theo Let us know about the most annoying bugs in Air, we’ll try to address asap 🫶
English
Kind of wild that the best agentic coding experience available today is an open source app made by some youtuber
English
@getsome_air 2. Feature: Just like with Agent Review it would be nice to also be able to customize Commit Message Agent pick its model and it spawn its subagent.
3. Feature: Manage models - A way to favourite or pick few models that show up in the chat options
English
@getsome_air Just got it working and i'm loving it hope to see it improve.
Two thoughts though
1. Bug: while in plan mode answering agents questions in the popup it closed before i answered all the questions
English
@opencode @metruzanca Even after turning it on in settings every time i open OC desktop the first chat doesnt have Plan/Build i have to open a second one then go back and close the first one
English
@metruzanca You can turn this on in the settings but with the newer models you don't typically need a separate plan mode
English
@ninopiamonte @opencode Free endpoint is 256k and paid is 1M context window! Have fun!
English
@iamdavidhill @sudoPhoeniX @opencode On the desktop app can Build mode and Plan mode have different colors I keep planing in build mode cause it looks the same unlike in the terminal 🙏
English
@JohnSM_99 @Spshulem Soon enough it will be much much cheaper. But the idea is the same.
English
“So you spent $40K to run GLM 5.2 locally?”
“Yes, Dave”
“And you did all this to save $20 a month on Cursor”
“Thats right, Dave”
English
English
I love that every time I open Codex CLI it tells me that GPT-5.5 is now available 🙃
English
This is a real thing.
I have it even before I start now. If something is long and tedious and hard to understand, my whole body resists the task. And I start looking for other things to do that just involve prompting.
I do feel like I'm being trained to stop thinking.
vicki@vboykis
Something happens to my brain after agentic coding that I can’t describe. It’s like cognitive offloading which folks have already written about, but even more. It feels like I can’t think through problems anymore. Like a fog. Using agentic but losing my hard-won agency.
English
@burkeholland completely agree right now everyone is just walling their garden with tools that only work well with their models
English
It seems to me that significant improvements in models won’t happen until we discover a method other than simply increasing the number of parameters.
This is why it feels like we kinda hit a ceiling with Opus 4.5. Cause we did.
English