Founder of Tamdoma. Growth marketer & AI transformation/automation Engineer. Christian.
Zagreb
Joined January 2010
- Tweets2.2K
- Following1K
- Followers260
- Likes1.1K
Ivan Juras retweeted
Day 25/
We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today.
We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage.
And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
Ivan Juras retweeted
Introducing Gemini 4 Carbon, it's a big jump over Argon which we released just 2 weeks ago.
Gemini 4 Carbon is SOTA on cyber, long horizon tasks, ranks #1 on Artificial Analysis V5, significantly improves on usability in domains like coding, and is more aligned.
Anthropic did not burn $80.7B in six months.
$72B of that is a paper charge on convertible notes, which are loans investors can later swap for shares.
Simple version: you borrow $100 to open a bakery. Your lender can take the $100 back or 10% of the bakery. The bakery becomes worth $10,000. That 10% is now worth $1,000, so accounting books a $900 loss. Nothing left your bank account. You lost on paper because the bakery got more valuable.
Strip the charge out and H1 2026 is:
Revenue $16.3B
Loss ~$8.7B
In 2025 Anthropic lost about $1.76 for every $1 of revenue. In H1 2026 it was about $0.53.
Still a real loss. The number worth watching is whether revenue outruns a $600B+ compute bill, not the $80B headline.
Ivan Juras retweeted
🚀Excited to announce Nanbeige4.3-33B!
To push the performance ceiling of small agents, our strategy in Nanbeige4.3 starts with building a powerful teacher first.
Nanbeige4.3-33B delivers SOTA agentic performance among open models at ~30B parameters:
* Strong results on agentic coding and cowork benchmarks.
* When integrated into mainstream agent harnesses, it tackles real-world tasks with cost-efficient inference.
* It will serve as a teacher for upcoming smaller Nanbeige models, starting with Nanbeige4.3-3B.
huggingface.co/Nanbeige/Nanb…
huggingface.co/Nanbeige/Nanb…
#LLM #Agent
Ivan Juras retweeted
"We figured out how to train Dessa on only one NVIDIA GB200 NVL72"
I'm blown away! Dessa is WOW!
Now this is something that can be used to create realistic customer support voice agents.
Meet Dessa, our speech model trained from scratch on human conversation.
Hear it for yourself: block.xyz/dessa
Ivan Juras retweeted
EAGLE-4 is here.
From EAGLE-1 to EAGLE-3, we’ve kept making inference faster.
EAGLE-4 is our fastest yet.
961.6 tokens/s on @Alibaba_Qwen’s Qwen3.8-27B with TokenSpeed (BF16, 4× @NVIDIA B300, TP4, one request, T=0).
Github: github.com/eagle-hq/EAGLE-4
Ivan Juras retweeted
Omarchy 4.5 RC1 is available as a pre-release ISO for both x86 and ARM, as well as on the RC channel. Absolutely packed with features and improvements. Please help us test it! github.com/omacom/omarchy/re…
swe-2-max is such a workhorse. I hope they keep it unlimited. It's so good. There's no way I'd be able to do as much even with 5 Claude subs (I have 3-4 all the time).
Ivan Juras retweeted
Day 16/
You can now cache for 12h in the API and for logged in users across Codex & Work we keep cache forever and you don’t have any penalty for coming back to an old conversation or resuming after a few hours and are now guaranteed to hit the cache.
Ivan Juras retweeted
Omarchy v4.0.5 release is here!
This release brings a bunch of fixes and refinements to tighten things up as we prep for the next feature release coming soon. It also bumps the packages on the stable ring, and brings an updated Omarchy kernel with BORE scheduler.
github.com/omacom/omarchy/re…
Ivan Juras retweeted
anthropic made claude code 3x faster using gzip.
a compression tool from 1992.
Ivan Juras retweeted
🚀 DeepSeek-V4.1-Flash-1017 is here!
Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API.
🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model.
1/4
Ivan Juras retweeted
“Claude writes about 90% of Claude Code's code.”
the absolute legend writing the other 10%:
Claude writes about 90% of Claude Code's code.
From June to September, merged PRs more than doubled while the share reverted fell by more than 40%.
We wrote about how we changed code review to get there, and how to adopt it:
claude.dev/blog/solving-code…
