08:18
Qwopus 35B + MTP: The Coder That Fixes Its Own Bugs at 160 tok/s
Fahd Mirza tests Qwopus Coder, a 35-billion-parameter mixture-of-experts coding model built on the Qwen 3.6 architecture (3B paramete...
Sort by: Newest Items
There are 70 items in this page
25:57
I benchmarked the NEW Sonnet 5. The results shocked me.
How I AI introduces the Howi AI Bench — a repeatable, multi-dimensional evaluation framework built with Claude Code — and runs Claude...
30:52
Frontier results, on device – RL Nabors, Arize
Rachel Lee Nabors — formerly at Mozilla on Firefox DevTools, the W3C, Microsoft Edge, and the React team, now at Arize — presents a p...
13:57
Can Krea 2 Turbo Really Make Great Images in 8 Steps? ComfyUI Test
Veteran AI runs a structured eight-category evaluation of Krea 2 Turbo — the eight-step distilled image generation model released by...
14:08
Qwythos 9B: When You Train a Small Model on Claude Traces: Run Locally
Fahd Mirza introduces and benchmarks Qwythos 9B, a reasoning-focused open-source model fine-tuned on over 500 million tokens of Claud...
09:36
Qwen3.6 (REAP 90pct GGUF): The Brain-Damaged Model
Fahd Mirza takes a deep look at an aggressively pruned variant of Qwen 3.6 — a 35-billion-parameter mixture-of-experts model — compre...
18:17
VibeThinker 3B – Taking on Giant Models
Sam Witteveen digs into VibeThinker 3B, a small language model from Waybo AI Lab — the AI research arm of the Chinese social network...
08:20
LoopCoder – The 7B Model That Thinks Twice – Does it Beat Others?
LoopCoder V2 is a 7-billion-parameter open-source code model built on an unusual architectural idea: instead of stacking more transfo...
09:40
DFlash Just Got Faster: 4x Speed with 160 tok/s Locally
Fahd Mirza benchmarks DFlash with SGLang's new SpecV2 overlapping scheduler on an NVIDIA H100 80GB GPU, demonstrating a 4.3x throughp...
31:25
Claude Fable 5 BANNED: The First Model Agentic Engineers DON’T NEED
IndyDevDan covers two intertwined stories in this video: the sudden federal suspension of Claude Fable 5 and Mythos 5, and a detailed...
09:03
I tried to prove AI trading is BS and it backfired
The Algovibes channel set out to definitively disprove AI-powered crypto trading — and ended up with results more interesting than ex...
09:13
I Tested 100,000 Trading Strategies on 1,000 Stocks
Algovibes presents a large-scale systematic backtesting study covering the full Russell 1000 universe — 1,014 stocks, 66 technical tr...
14:35
Google QAT vs Unsloth Q4_0 – Which Gemma 4 12B Quantization Is Better?
Fahd Mirza runs a controlled comparison between two 4-bit quantized versions of Google's Gemma 4 12B model: Google's own QAT (quantiz...
12:30
Ideogram 4: World’s Best Text-to-Image Model? Let’s Test Locally
Fahd Mirza installs and tests Ideogram 4 locally, providing a candid assessment of its real-world hardware requirements and architect...
14:55
Gemma 4 12B on a 16GB Mac Mini Is Surprisingly Capable
Bart Slodyczka puts Google's newly released Gemma 4 12B model through its paces on a 16GB M4 Mac Mini — a practical test of what entr...
16:08
Benchmarking semantic code retrieval on Claude Code — Kuba Rogut, Turbopuffer
Kuba Rogut of Turbopuffer presents original benchmark results comparing three code retrieval strategies for Claude Code: the default...