NOKOL

GER 34y | Fullstack Developer & Architect

[NOKOL @Nokolig]

Ok GPT 6 esp. Astra are fucking bad now.
Tested it broadly. Sucks bad

[Tibo @thsottiaux]

We're going beyond text and are releasing a new version of GPT-6 to all users in ChatGPT. Lots of model and infra improvements coming together here to scale it to 1.2B users!

[NOKOL @Nokolig]

People don't know Aleph Alpha made a LLM 2024 which was shit - raised capital and changed their direction

Now they come back "hyping"nwith sth from scratch.

I'm afraid but curious

[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.

[Abhishek Singh @0xlelouch_]

90% of backend engineering in 2026 comes down to mastering these 10 concepts:

1) Concurrency + backpressure
Bound every queue, thread pool, and consumer; dropping work is often better than melting the DB.

2) Data modeling + invariants
Write down what must never break (unique, monotonic, balanced ledger) and enforce it in the database, not in vibes.

3) Idempotency + retries
Every network call will retry; use idempotency keys and dedupe so at-least-once delivery does not become double-charging.

4) Consistency tradeoffs
Know where you need read-your-writes vs eventual; most outages are mismatched expectations, not bad code.

5) Cache correctness
TTL is not a strategy; handle stampedes, negative caching, and invalidation when the source of truth changes.

6) Deploy safety
Small rollouts, fast rollback, and feature flags beat hero debugging; schema changes need a zero-downtime plan.

7) Observability that answers questions
Golden signals plus high-cardinality traces tied to a request ID; logs without context are just expensive text.

8) Production debugging
Learn to prove where time went: CPU vs IO vs lock vs network; p95 is usually one hot key or one slow dependency.

9) Security basics that bite
AuthZ checks at every boundary, least-privilege IAM, secrets rotation, and input validation that assumes hostile clients.

10) Tooling and runbooks
Repeatable local repro (docker, seed data), one-command deploy, and a runbook that says what to check at 2am (dashboards, queries, kill switches)

[Jason Zhou @jasonzhou1993]

15min @typesafe Jeveloper tutorial 🧠

My team at @treg_ai has been using Jev a lot in prod:
- Classify web traffic
- Smart onboarding
- Jev-reranked search

~10x cheaper, 18x faster, 30% more accurate in our tests compared with GPT6-luna

OpenSourced our Jev implementation 👇 https://t.co/Qypv2lQv1u

[I Am Devloper @iamdevloper]

TDD became *almost* obsolete once LLMs starting writing/reading most of the code

Heavily specced integration tests > unit tests

[Jake Fitzgerald @earthtojake]

text-to-cad plugin extension is now live on codex!

- generate 3D models as STEP, STL, 3MF, or GLB
- DFM for printing, sheet metal, CNC, injection moulding
- connect to fab services like bambu and sendcutsend

100% open source and free, runs locally inside the codex desktop app

[Stefan 3D AI @Stefan_3D_AI]

Free Local Image-to-3D Now Keeps Your Text and Logos Readable, Even at 5K Faces

Image to 3D Lab turns a single image into a textured 3D model on your own machine. The new Pixel Match update copies the real pixels from your picture back onto the model.

Highlights:
• Text and logos stay readable, even at ~5K faces (down from ~900K)
• Low-poly models you can rig and animate
• Runs locally on Mac or NVIDIA
• Free and open source, no API or credits

GitHub: https://t.co/NFDgTMD283

[Blendi @BlendiByl]

I open-sourced Dioramas, a framework for cinematic, interactive 3D websites.

It's 100% free. No paid tier, no signup. It comes with 20 full example sites, each built around one interaction you can touch: a museum you explore by lantern, a mech you power up, a geode you crack open.

It's powered by fal: every 3D model was generated with Meshy 7.1 from Nano Banana 2 images. Every prompt, model and line of code is in the repo.

Repo: https://t.co/ZsSXaVgU0u

[Hang Huang ☁️ @hanghuang_]

We just raised an $8M seed round to kill AWS, GCP, and Azure.

Introducing https://t.co/E3LHVE410x, the agent-native serverless cloud.

Your team is shipping code like never before. But you're getting caught up in manual, tedious DevOps work trying to deploy it.

InstaCloud provides the serverless compute that lets your services autoscale, with all the infrastructure managed for you.

Agents branch into complete replica environments when working, keeping prod safe and iteration speed high.

And of course, it all works seamlessly with agents through MCP/CLI.

Get off the traditional, legacy cloud.

Start deploying your services on InstaCloud today.

[Lance Martin @RLanceMartin]

Claude Code can help with eval design + hillclimbing.

check out this article i just published for details:
https://t.co/fg3VvVwwpk

[NOKOL @Nokolig]

@hwchase17 Repo harness - department harness - company harness

[NOKOL @Nokolig]

@petergyang Wtf over

[Yafeng Deng @LongTermMemoryE]

Raven 0.2.0 — The Harness of Harnesses, built for RSI. 🐦‍⬛

One harness can't be best at everything. Raven combines its own specialist harnesses (Research, Code, Design, Oncall) with the agents you already use (Claude Code, Codex and more) into one team.

And it's built for RSI, and not just at the skill level. The whole harness can be rewritten by AI: prompts, policies, strategy code, playbooks. Every sub-harness, including the orchestration layer itself, is its own instance that can be improved.

With Raven you can:

1. Orchestrate many agents as one team. Raven's sub-harnesses and external agents work in one task graph with shared memory across sub-agents, powered by leading orchestration (0.963 Node F1 on the Multi-Agent Orchestration Benchmark).

2. Run long, complex tasks. Oncall and proactive execution keep work going for days, from scientific research loops to shipping a full Godot game.

3. Build vertical agents with RSI. Use Raven's RSI to develop and refine an agent for your domain, and we'll optimize it with you. Experimental for now; reach out to the Raven team(Discord:https://t.co/jRrci3hVL5).

More in the video and slides below. Open source, Apache-2.0.

https://t.co/2GEA4Nmig0 (lots of work made with Raven lives there, and much of this launch's material was made with Raven too)

[Polymarket @Polymarket]

JUST IN: New study reveals creatine may increase muscle mass, strength, cognitive function, & endurance even without exercise.

[Wenhao Chai @ COLM @wenhaocha1]

Can we predict how agent swarms will behave without burning thousands of dollars💰 on experiments?

We built a simulator that generates task DAGs (from real RSI experiments) and lets swarms re-play with them: a cheap way to study swarm scaling before paying for the real thing.

[NOKOL @Nokolig]

Gpt-6-sol is super bad. Like wtf

[Kit Langton @kitlangton]

jesus one-shot christ

[Yiwei Ho @1weiho]

Introducing open-slide 2.0 🎉

The slide framework built for agents, now with:
› A new visual editor
› Editable pptx export
› A redesigned UI

Let your agent build the deck. Make the final touches yourself.

Here's what's new ↓

[NOKOL @Nokolig]

@l3d1c I'm 15y dev and ranked as junior
60k salary working for a big as company with a lot of money.

It's my fault

[Kieran Klaassen @kieranklaassen]

How to use jev #1: As an embedder