Omar Sultan

@emanueledpt Can I ask u something, do I loose any capabilities in using Claude Code in Synara vs using it through the cli? Does it rely on the same harness ?

I just used Claude Opus 5 in Synara to do a huge refactor + performance boost in Synara itself So I asked Opus to create a chat in Synara with 5.6 Sol xhigh fast to review the work And I'm using Synara Automations to check the work to tell me then what it found Synara is AGI

@emanueledpt I love it , but my problem is that it doesn't use Claude Code but rather the Agent SDK :(

People want solutions But not my solution This is the game where you’re not famous enough GG

@anur7ag i want this

what if herdr had a desktop app?

@matt_feroz The ui is perfect , but can we support terminal please

This needs to be the new design paradigm for coding with agents. I'm trying my best not to glaze here, but compared to what it was previously this UI is close to perfect for me

@leodev What about codex vs Claude code as the harness , have you noticed a difference

I'm cancelling Claude because the limits suck... But the main reason is that GPT 5.6 doesn't suck with UI that bad anymore. Still not Fable level but it's 90% of the way there. I'd rather use it with insanely high limits compared to Fable.

@kunchenguid I have noticed the spiraling out of control part, sometimes things that doesn’t require much end spawning endless agents I’ve also had an incident when it kept trying to fix some database migration issue and kept resetting the dev database and then went in an endless looo

after a few more days of using gpt 5.6 sol, i started noticing some issues - if you have good solutions, please share! 1. it uses technical jargons a lot. it's almost speaking its own language that looks like English but you can't understand it until you ask "what do you really mean by this" 2. it can over-engineer and spiral out of control. something that can be done with a few lines of changes can often result in a massive diff fixing everything in the codebase 3. it's overly conservative in terms of touching live environment, so much so that it overly relied mocks for development and validation and build things that don't really work in production these things can likely get tweaked in system prompts, but so far gpt 5.6 is the only model family that don't get these things right out of the box, and it's a bit annoying i love overall how intelligent and fast the model is though! just curious if anyone's run into similar challenges with the model and has good solutions

@thsottiaux Give us a banked reset ;)

Embarrassment of riches. But looks like we might hit 9M soon. Should we reset the ChatGPT Work and Codex usage again or give it some space?

@thsottiaux @OpenAI The resets , the writing capability of the model in terms of documents just feels much more natural , switched all my writing to it 🥳🥳🥳

Or… what if we gave you $100 in Codex credits if you tell us what you love about GPT-5.6 Sol or why you switched? Tweet it, claim your gift, enjoy more usage. First 10k get the free tokens! switch-to-codex.openai.chatgpt.site

@thsottiaux You are making it kind of hard to go back to Claude code :)) loving the resets

Hello. We have reached 8M active users across Codex and ChatGPT Work. We are once again resetting the usage limits for all. And we continue to not have the 5h rate limit as well, allowing everyone to explore the boundaries of GPT-5.6 Sol and discover how ambitious you can be. See you tomorrow for more updates on our growth!

@trySynara I love the app , but my problem is having to work with Agent SDK rather than Claude Code in terminal , how can we work with Claude Code in terminal in Synara

Introducing Synara 0.5.0 — the new Synara 🪽 This release marks the beginning of a new chapter. Synara now has its own identity across the entire product. No more legacy T3Code or t3tools references. From the interface to the desktop app, everything is now unmistakably Synara. Synara is now Synara, all the way through. Please make sure to download the latest version.

@rezoundous will probably downgrade to 5x or something , use Opus for some frontend stuff but use gpt5.6 using Proxy or sth on Claude Code

Would you continue to stay subscribed to Claude if Fable gets removed?

@chrisrismay @OpenAI me too :(

@OpenAI Is 5.6 out? I dont see it.

Introducing ChatGPT Work, a new agent in ChatGPT powered by Codex and GPT-5.6. It can take action across your apps and files, stay with a project for hours if needed, and turn a goal into finished work. It’s a whole new way to get work done.

@skirano How does it compare to Fable in terms of writing skills if we’re talking about writing documents that are mostly technical?

I can finally talk about 5.6. I’ve been testing it for months and, without exaggeration, it’s the best model I’ve ever used. Fast, smart, genuinely creative, and you guessed it, they finally fixed front-end design. I haven’t needed to check the code I’ve written in two months.

@herdrdev its a brilliant app :) love it

herdr is 105 days old. i started writing a 10,000 github stars celebration post. before i finished it, we passed 12,500! on june 30, herdr hit #1 on github trending. it has stayed in the top 10 since. i’ve been building it solo and full-time, in public from day one: roadmap, issues, feature requests, releases. so instead of just celebrating, i’m opening the stats too: stars, downloads, website traffic, releases, and where the numbers come from. herdr.dev/stats thank you for being part of the herd 🐑

@kurtbuhler Thank you so much for this

1. Get an agent like Claude code, Codex, or GitHub Copilot CLI. 2. Save your report in the PBIP format with PBIR metadata 3. Install the pbir CLI with uv install pbir-cli. If you don't have uv install with winget uv. You do this in the terminal. 4. Start the agent and ask it to install the pbir-cli skill from data-goblin/power-bi-agentic-development on GitHub 5. Work with the agent to modify the report. Ask it to log learnings as it works.

I've created a benchmark that measures agent performance in basic Power BI report tasks, showing that directly modifying PBIR files is worse, more expensive, and (often) slower than using pbir-cli. The tests evaluate 30 simple tasks like "create a card that shows the total invoice lines, using " and a pass is when the agent can do it correct across 3/3 replicates. So if it does it 1 or 2 times, it still fails. The results were quite interesting. A plugin with skills and small CLIs for validation didn't meaningfully increase performance much (esp not for Opus 4.8), but ballooned token cost and time. In contrast, the pbir-cli alone was sufficient to increase performance significantly on those tasks, and adding the skill made it faster and take less tokens. "Dumber" models took longer with the CLI because they spent more time exploring or iterating, and seem worse at using skills in general. Obviously together with @maxanatsko I'm an author of the pbir-cli so it's appropriate to be skeptical of my benchmark that favors my own tool, but I did try to make these tests objective and fair. Note this benchmark only evaluates performance as a function of "can it complete the task" not "do the report / visuals look nice". I'm building a design evaluation as well but it takes a lot longer since design is subjective and coming up with good, repeatable tests that aren't handwavy marketing bs (like "make the report from this image") is challenging.

@iam_daniiell What about the reports themselves ? Did Claude build that too?

𝗜 𝗱𝗶𝗱𝗻’𝘁 𝘄𝗿𝗶𝘁𝗲 𝗮 𝘀𝗶𝗻𝗴𝗹𝗲 𝗗𝗔𝗫 𝗺𝗲𝗮𝘀𝘂𝗿𝗲. 𝗖𝗹𝗮𝘂𝗱𝗲 𝗱𝗶𝗱. The Operations Director knew something was wrong. Costs were rising. Deliveries were slipping. Packages were disappearing. Every time the executive team asked why, there was no answer. Not because the data didn’t exist. Because nobody had built the thing that made it visible. That was the brief. 2,000 shipments. 7 carriers. 10 warehouses. 13 states. We built a 3-page dashboard that put the answer in front of leadership before the meeting started. 𝗨𝗣𝗦 leads the network at 86.3% on-time. 𝗔𝗺𝗮𝘇𝗼𝗻 𝗟𝗼𝗴𝗶𝘀𝘁𝗶𝗰𝘀 sits at 79.2% with the worst problem rate. One scatter plot answered the question every operations manager is quietly asking: Are we paying for performance, or just paying? 45 lost shipments. 32 returned. 14 cases still open. Visible the moment you open the dashboard. Built this live with a student. Every DAX formula was written by Claude. My job was the thinking. The model. The layout. The business questions each page needed to answer. 𝗔𝗜 𝗰𝗮𝗻 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗲 𝗺𝗲𝗮𝘀𝘂𝗿𝗲𝘀. 𝗜𝘁 𝗰𝗮𝗻’𝘁 𝗱𝗲𝗰𝗶𝗱𝗲 𝘄𝗵𝗮𝘁 𝗺𝗮𝘁𝘁𝗲𝗿𝘀. That’s still the analyst’s job. Explore it: app.powerbi.com/view?r=eyJrIjo… Open it, interact with it, there’s more in it than you think. Tell me what you think. #DataFam #PowerBI

@blackwellboy @UnslothAI M5 Max 128gb

Drop your GPU below and I’ll tell you exactly what model and config to run on it. JOKES. No need. Qwen 3.6 27b @UnslothAI . All day. All night.

@kurtbuhler i really wanna work with it but for some reason im having a hard time getting my head around it , do u have some guides or videos on how to use it?

Really happy to hear people using the PBIR CLI - Tableau --> Power BI migrations - Mass fixing of broken reports - Report CICD with BPA - Bulk changes of formatting or fields Some of the use cases I've heard people share in the last week on comments or DMs. More coming!

@LinusEkenstam Created this Agentic Task Managemetn System sentratasks-ai.com

Can I see some links of what you’ve built with Claude code? Blow my mind