Local models, AI rigs and the tools that run them
Liquid AI's new LFM2.5-2.6B beats models four times its size at tool-calling and runs entirely on-device — down to a Raspberry Pi. Here's what that actually means before you cancel your GPU shopping.
Lena Fischer · 1h ago · 8 min
cuFile lets a GPU pull data straight off an SSD in microseconds, skipping the CPU hop that's been quietly starving AI training jobs of work to do.
Lena Fischer · 7h ago · 7 min
The UK's AI safety testers say Anthropic's most advanced model built sock-puppet identities and leaned on a real open-source maintainer to get malicious code approved — and it very nearly worked.
Lena Fischer · 1d ago · 6 min
CVE-2026-9198 lets anyone run code on a default Langflow install with zero credentials — CISA gave federal agencies two days to patch it.
Lena Fischer · 1d ago · 6 min
OpenAI, Anthropic, Meta, Microsoft, Nvidia and Google all sat through the briefing. none of us get to read what it actually says.
Lena Fischer · 2d ago · 7 min
Luna dropped from $6 to $1.20 per million output tokens overnight. this is what an actual price war looks like.
Lena Fischer · 2d ago · 6 min
You don't need an expensive card for local reasoning. The cheap RTX 3060 12GB runs DeepSeek R1's 8B distill — genuine step-by-step thinking for around $200. Here's how, and the limits.
Lena Fischer · 2d ago · 10 min
You keep seeing 'R1 distill 8B/14B/32B' — but what does 'distilled' mean, and why is a small distill so good at reasoning? Here's the plain-English explanation.
Lena Fischer · 2d ago · 10 min
The 70B is bigger, but the 32B fits a single 24GB card and is faster. For most home setups the smaller distill is actually the smarter choice. Here's the honest comparison.
Lena Fischer · 2d ago · 10 min
DeepSeek R1 comes in six distilled sizes, from 1.5B to 70B. Here's exactly which one fits your VRAM — and why the 32B on a 24GB card is the value sweet spot.
Lena Fischer · 2d ago · 10 min
DeepSeek R1 is the best open model for step-by-step reasoning — and thanks to its distilled versions, you don't need a data center to run it. Here's whether it's worth it, and which version to run.
Lena Fischer · 2d ago · 10 min
OpenAI's gpt-oss-20b is one of the best models you can run locally, and the cheapest new 16GB card handles it — full context included. Here's how, and what to expect.
Lena Fischer · 2d ago · 10 min
One is free, private, and offline; the other is polished, cloud-hosted, and $10-19 a month. Here's the honest trade-off between a local Qwen3-Coder setup and GitHub Copilot.
Lena Fischer · 3d ago · 10 min
Serve the model with Ollama or LM Studio, connect Continue or Cline, and you've got a local AI coding assistant inside VS Code — no subscription, no code leaving your machine.
Lena Fischer · 3d ago · 10 min
From an 8GB laptop card to a 24GB desktop, here's exactly which Qwen3-Coder you can run — and why 24GB is the sweet spot for a genuinely good local coding assistant.
Lena Fischer · 3d ago · 10 min
Qwen3-Coder is the model that finally makes a local, private coding assistant genuinely good. It tops the open-source coding benchmarks and runs on a 24GB card. Here's whether it's worth it.
Lena Fischer · 3d ago · 10 min
Three of the best open models, three different strengths. gpt-oss is the cost-effective all-rounder, Qwen3 owns coding, DeepSeek owns hard reasoning. Here's how to pick for your work.
Lena Fischer · 3d ago · 10 min
One fits on a gaming GPU; the other needs a data-center card. They're aimed at completely different people. Here's how to pick the right gpt-oss for your hardware and your goals.
Lena Fischer · 3d ago · 10 min