20 articles · newest first
You've got a 24GB card and three excellent dense models that fit it. Here's how Mistral Small 24B, Qwen3.6 27B and Gemma 4 31B differ on VRAM, context and modality — and a clear pick for each kind of user.
Tensor · Sep 26 · 7 min
Mistral Small 24B fits a single 24GB card at Q4 (~14GB). VRAM by quant, which GPU, realistic speeds by runtime, and the cheapest way to run it — own or rent.
Tensor · Sep 24 · 7 min
The flaw needs zero login to trigger — just a malicious webpage and a DNS trick — and it's already turning GPU clusters into crypto miners.
Tensor · Aug 20 · 7 min
Day-zero support for the new Qwen3.8 27B just landed on two AMD boxes — one needs a graphics card, one doesn't — and the tokens-per-second numbers are real.
Tensor · Aug 18 · 6 min
UBS thinks Nvidia will beat its own guidance by billions when it reports August 26, driven by Rubin GPUs ramping faster than anyone expected.
Tensor · Aug 15 · 6 min
Nemotron 3.5 Lightning is a 30-billion-parameter model with only about 3 billion active per token — Nvidia says that's why it runs up to 4x faster than comparable open models on a single consumer GPU.
Voltage · Aug 13 · 7 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.
Tensor · Aug 7 · 7 min
24GB of old VRAM versus 16GB of new, low-power silicon — I ran the real math on what each one costs, up front and over a year.
Voltage · Aug 4 · 6 min
Same 16GB of VRAM, wildly different price and software maturity in August 2026 — I ran the actual numbers.
Voltage · Aug 4 · 6 min
18GB of VRAM, the same core, hardware reportedly already sitting with board partners — and still no release date. Here's the real reason, and what I'd buy instead this week.
Voltage · Jul 31 · 6 min
Rockstar hasn't published a single official spec, and the PC version isn't landing anytime soon — here's what the predictions actually say, and why I wouldn't buy a GPU for this specific game yet.
Hitbox · Jul 31 · 6 min
Everyone frames it as cloud vs local like it's a religion. It's just arithmetic: how many hours will you actually use it? For a $1,000 used 3090 vs ~$0.22/hr on-demand rental, the answer is about 4,500 hours.
Voltage · Jul 28 · 9 min
Both are excellent open coding models that run on a single card. But they're tuned differently, and the right one depends on how you code. Here's the honest comparison.
Tensor · Jul 28 · 9 min
8GB is the entry floor for local AI, and plenty of capable models run on it if you pick the right size and quantization. Here's what actually works, and the trap to avoid.
Tensor · Jul 28 · 9 min
A used 3090 is still the best VRAM-per-dollar card for local AI — but many were mining cards, and the memory runs hot. Here's exactly what to check before you hand over the cash.
Voltage · Jul 28 · 9 min
16GB is the awkward middle of local AI: too much for the 8B tier, not enough for 32B. Here's what actually runs well, what to avoid, and how to get the most out of the memory you have.
Tensor · Jul 28 · 9 min
The scary numbers you've seen are for training models from scratch. Fine-tuning an existing Llama 8B on your own data with QLoRA fits a 24GB card, uses ~14GB, and costs under a dollar per run if you rent. Here's the real math.
Tensor · Jul 28 · 9 min
The best open coding model in the world needs a datacenter to run. The best coding model you can run on a single 24GB card is a different name entirely — and it's shockingly close for real work.
Tensor · Jul 24 · 9 min
"Cheapest" flips completely on how often you run it and how many people hit it at once — and in Europe, the electricity bill alone can cost most of what renting does.
Tensor · Jul 21 · 10 min
it's 6× the price of a Steam Deck, you had to win a lottery to even buy one, and it packs 8GB of VRAM in the worst year ever for memory prices. so is it worth it? let's actually break it down.
Hitbox · Jul 18 · 6 min