Nils Ingwersen
267 posts
Nils Ingwersen
@nilsingwersen
Dev @ubilabs
Hamburg, Deutschland Katılım Haziran 2013
341 Takip Edilen35 Takipçiler
Today we’re releasing DeepSWE, a new standard for agentic coding benchmarks.
On public leaderboards, top models often look relatively close in capability. DeepSWE shows where they actually diverge, reflecting the realistic experience of developers in their day-to-day work.
English
@schickling Definitely. We went to the cinema in shinjuku and everyone watching the credits while being quiet was so nice to see.
English
Coming back from a 3 week Japan trip to Germany. The reverse culture shock hits hard. 🫠
English
@lubinho_k Because Kimi the base model is really fast. And with post training the capabilities become very good. But the new gemini 3.5 is also super fast. Codex for me also very fast when running in pi.
But yeah running kimi on fireworks ai is absurdly fast.
English
Been working with Composer 2.5 all morning. Damn the pace this guy moves at is just amazing. Really getting used to it and really hard to go back from.
Can anyone explain to me why it's so much faster than Codex and Opus while delivering similar outputs?
English
@googleaidevs @antigravity You converted me to a paid plan. Initial reactions had me skeptical but your comms and tripling the plans twice got me.
Keep up the good comms.
English
Update: *Weekly* Gemini quotas on all paid @Antigravity tiers have just been tripled (again!), and quotas are officially reset (again!).
Varun Mohan@_mohansolo
Yesterday, we 3x’d limits on Antigravity and are seeing you build so much more. One thing we heard was people are worried about hitting their weekly limits after a couple work sessions. To give you more runway, we’re 3x’ing the weekly Gemini quotas AGAIN on all paid plans. We’ve also gone ahead and reset Gemini quotas on all paid plans. Don’t stop building!
English
@ShiroLinHime You can run /permissions and set it don’t ask for permission. But yeah the ctrl y shortcut for yolo mode was better
English
4. Random freezes with no way to exit.
5. Non-CLI mindset for plans/artifacts—makes planning & execution tedious.
6. Couldn't find a YOLO mode.
7. I'd list more flaws, but your stingy quotas ran out before I could even use it. 😮💨
I missed gemini-cli. It was way better.
English
My honest feedback on antigravity-cli:
0. Requires re-auth every launch despite being logged into the IDE.
1. Unreadable UI rendering during execution.
2. Zero UI hints for costs, skills, MCP, context, or current paths.
3. No auto-model allocation; manual switching is clunky.
↓
English
@_mohansolo Are you the person we can ask for resets like tibo?
English
An update: we’re 3xing the rate limits for Gemini models across all paid tiers in Antigravity and resetting everyone’s Gemini quota for the week.
We understand some people hit their rate limits quickly and wanted to respond fast. Lots more to come and enjoy building!
English
@nicodotdev Have fun, I'll go in july there. Wish I could go to IO aswell.
English
On my way to San Francisco🤩.
First time in California. First time in Silicon Valley. And I'm going for Google I/O. 🤯
Also I recently started a new instagram channel. So if you want the unfiltered version, follow along👇😊
English
@TJ_Kawa The idea is cool but the map is not really usable on mobile. It keeps resetting the zoom.
English
Japan has 4M+ vending machines, called jihanki.
In Tokyo they’re everywhere, but the interesting ones are weirdly hard to find.
So I made Jihanki Atlas, a map where people can discover, submit, and help verify Tokyo vending machines.
Current favorites: Pokémon plushies, sriracha sauce, wagyu beef, edible insects, fresh-squeezed orange juice.
What’s the coolest one you’ve seen?
jihankiatlas.com
English
@AbroadInJapan Are you watching in cinemas or at home and are there any movies you are looking out for this year? Personally I can’t wait for Dune Part 3 although I’m a bit sad that Greg Fraser isn’t working on it.
English
One of my goals in 2026 is to watch a lot more films.
I've always been a lover of movies, but over the last few years I’ve realised I want to be more intentional about what I watch and what actually sticks with me afterwards.
Kicked off a profile on Letterboxd to keep track, exchange the odd thought and slowly build a record of the stories, performances and deranged third acts that stay lodged in my brain.
English
Recruits, your first prize is here...
A custom GeForce RTX 5080 Founders Edition + PC copy of the game.
Comment #007FirstLightRTX to win 👇
English
@theo Imagine you could use the claude agents command and get that output into your sidepannel
English
Good news: T3 Code is officially supported by Claude now
Bad news: it bills API credits 🙃
ClaudeDevs@ClaudeDevs
Starting June 15, paid Claude plans can claim a dedicated monthly credit for programmatic usage. The credit covers usage of: - Claude Agent SDK - claude -p - Claude Code GitHub Actions - Third-party apps built on the Agent SDK
English
@kr0der @FireworksAI_HQ has insane tps so that and then either kimi or deepseek
English
hypothetical scenario:
all the $20, $100, $200 AI coding subscriptions are gone, no more heavy subsidisations
what setup are you using now that you have to pay API prices?
English
@_philschmid Insane movie right? Any other anime you watch?
English
Just watched chainsaw man: reze arc after missing it in cinema. Great movie.
English
Interesting data on the next.js evals:
- GPT5.5 has the highest success rate by a tiny margin
- But it is >3x slower than 5.4
- And newer GPT models have been helped less by AGENTS.md instructions over the newer generations (Claude models usually pass at 100% with this help)
nextjs.org/evals
English
link to demo: flat.ingwersen.dev
docs: developers.google.com/maps/documenta…
English
I had some fun playing around with @GMapsPlatform new flat component and @fal for 3d model generation a while back. Really love flattening and godzilla.
English
Nils Ingwersen retweetledi
Rspack 2.0 has landed 🦀
The growth since 1.0 is wild:
📈 100K → 5M weekly downloads
⚡️ ~100% faster builds
📉 devServer: 192 deps → 1
And new stuff hits hard:
🧩 Pure ESM core, modern ESM library output
⚛️ Experimental RSC support
...
Lean, Fast, Modern Web Bundling 🚀
English