Niklas

CS student at TH Mittelhessen. Software engineer working on indoor navigation, web and Android apps, and machine learning. Striving for usable software.

[Amanda Goodall @Amanda_Goodall]

Well, this is awkward.

Back in March, I questioned the grand idea that AI agents could replace entire engineering teams.

Now Meta is reportedly rebuilding management layers because humans are spending too much time cleaning up after the machines.

Turns out firing the adults and leaving the interns in charge wasn’t much of a strategy…. even when the interns are artificial.

[Peter Gostev @petergostev]

Holy cow, so based on OpenAI's UK filings, average salary in 2025 was £794k and £1.52m share based compensation on top of that.

So the total was £2.35m = $3.1m per employee

[Niklas @nskrll]

@SebJohnsonUK @MistralAI @Aleph__Alpha @ElevenLabs Summer is clearly over 😉

[Mistral AI @MistralAI]

Meet Mistral Large 4, aka Le Chonk.

• 1T parameters, natively multimodal. 49B active.
It is the best open weights model from US or Europe on aggregated benchmarks.
• State-of-the-art on critical workloads, including cyber defense, manufacturing and finance and it surpasses closed frontier models on visual grounding.
• Forged in Europe end-to-end and is deployable from Europe via our own Mistral Cloud infrastructure.
• Available to all via API today. Working with cybersecurity partners privately.

Open weights release end of October.

[Aleph Alpha @Aleph__Alpha]

Small bird, fast wings, Kolibri is here.

78B parameters. 3.46B active. Up to 1M tokens of context. Built in Europe.

Now the weights are yours. Run it on your own hardware, under Apache 2.0.

[reed @reed]

second european @waymo city announced 🇩🇪

[Shuo Yang @Andy_ShuoYang]

Your gaming PC can now serve frontier models at interactive speed using official checkpoints without extreme quantization!

Qwen3.6 35B → 8GB RTX 4060 laptop @ 39 tok/s

DeepSeek-V4-Flash 284B → RTX 5090 desktop @ 22-25 tok/s

GLM-5.2 753B → RTX PRO 6000 workstation @ 15 tok/s

Run your claude code or codex now with frontier model for $0

Meet FreeToken 🧵

[Zander Whitehurst @zander_supafast]

Stop centering text in your UI

[Will Eastcott @willeastcott]

Google Maps helps you get from A to B. SuperSplat makes you feel like you're actually there.

Built on 3D Gaussian splatting tech from @playcanvas, photorealistic captures of real-world places stream to your browser in seconds. The best bit? It's all free and open source!
[1/2]

[Ole Lehmann @itsolelehmann]

should be obvious by now, but OpenAI and Anthropic are just gonna keep cannibalizing all their biggest customers.

it’s simply too profitable for them to resist. and it’s already happening:

1. Figma partnered with Anthropic on AI design tools. then Anthropic’s product chief quit Figma’s board, and 3 days later Anthropic launched Claude Design to compete with Figma. CEO Dylan Field said Anthropic was “not consistently candid.”

2. Novo Nordisk uses Claude to help develop drugs. now Anthropic is developing drugs of its own.

3. Microsoft poured billions into OpenAI. now OpenAI is building a Jobs Platform to compete with LinkedIn, and reportedly a code repository to compete with GitHub. Microsoft owns both.

4. Harvey uses Claude to sell AI contract analysis, due diligence, and litigation tools. now Anthropic sells those same workflows through Claude for Legal.

5. Intercom used OpenAI’s Realtime API to build Fin Voice. now OpenAI sells its own voice-and-chat support agent through Presence.

6. Abridge and Ambience build clinical documentation products on OpenAI. now OpenAI sells ChatGPT for Healthcare directly to hospitals with clinical documentation built in.

7. Benchling uses Claude to power its biotech R&D platform. now Anthropic sells its own scientific workbench through Claude Science.

the frontier lab playbook is simple:

1. sell their models to the world’s most valuable businesses
2. help wire them into those companies’ most valuable and sensitive work
3. map the business from the inside and find where AI can take over
4. turn those capabilities into their own products and become the customer’s competitor

[Niklas @nskrll]

@pmarca What are the key things in your opinion? You will never convince the bureaucrats and "we have always done it this way" politicians, but we, the voters, can bully these people until they give in or lose office if we know what to demand.

[Mike Matas @mike_matas]

After many years of development, I’m excited to share the interior of the first electric Ferrari designed by LoveFrom. Tactile controls and digital interactions blend into one cohesive interface, shaped through deep collaboration across engineering, interaction, graphics, typography, sound, and industrial design. So incredibly proud of the thoughtfulness and care the team brought to every detail.

https://t.co/JZCleflfu7

[sam henri gold @samhenrigold]

publishing my first app to the app store and was surprised to learn, in order to sell in the EU, I have to publish my home address. i get it for companies, but i am but a guy and my only address is the one where i keep my body and worldly possessions. i don’t wanna share it.

[Internal Tech Emails @TechEmails]

Microsoft CTO: Why the OpenAI board really fired Sam Altman

November 19, 2023

[Alvaro Cintas @dr_cintas]

"I don't have a GPU" is officially over.

VS Code now connects directly to Google Colab.

→ You get a free T4 GPU inside your editor.
→ Takes 2 minutes to set up. Their compute.

[Paul Couvert @itsPaulAi]

This is quite massive

Mistral has just released two open-source models including 'Devstral Small 2' which:

- Has only 24B parameters (28x smaller than DeepSeek 🤯)
- Can run LOCALLY on a laptop
- Is competitive with much larger models for coding

So basically everything you need for 100% privacy and unlimited vibe coding.

Plus, they're releasing a new vibe coding tool called Mistral Vibe CLI that you can use for free with their bigger model Devstral 2 (123B).

Western open source is so back.

[Aram Hăvărneanu @aramh]

1. People do check compiler output.

2. The reason why compilation is reliable is because programming languages are formal languages and we can prove (or at least try) that the output of a compiler is a semantic realization of its input.

The LLM crowd is delusional. LLMs are very impressive, but companies creating LLMs do not understand their own creations.

[Anthropic @AnthropicAI]

We disrupted a highly sophisticated AI-led espionage campaign.

The attack targeted large tech companies, financial institutions, chemical manufacturing companies, and government agencies. We assess with high confidence that the threat actor was a Chinese state-sponsored group.

[Rushil Shah @rushilshah_life]

Using the blur trick for optical alignment

[Evan Applegate @youwillmakemaps]

Day 2,368 of not shutting up about amber glowing maps (I started in 2019)