Insights for AI builders
Tutorials, product updates, and ideas to help you build and ship AI applications faster.
Subscribe via RSS
Nvidia eGPU on AMD Strix Halo: Oculink Setup and Real Speed Gains
How an Oculink port lets an Nvidia RTX GPU pair with AMD's Strix Halo APU in a mini PC, and what that actually does for local LLM speed.
Open Weights vs Closed Weights: Why Token Usage Is Flipping
Open-weight models now handle more AI tokens than closed models like GPT and Claude, yet closed labs still capture most of the revenue. Here's why.
Qwen3.8-Flash-Next: Inside the Qwen 4 Architecture Preview
Qwen3.8-Flash-Next previews Qwen 4's architecture: hybrid gated-delta and sparse attention, engram embeddings, 125B params, 6B active.
Retell AI Pricing, Free Tier, and How It Builds Voice Agents
How Retell AI's free tier, concurrency limits, and no-code builder work, based on a hands-on build of a phone-based AI voice agent.
How to Run Breeze TTS 2 Locally: GPU Requirements and Setup
Step-by-step guide to self-hosting Breeze TTS 2, covering GPU memory needs, Docker builds, and voice clone, design, and direction commands.
Run GLM 5.3 Flash Locally: VRAM, Quantization, and Hardware Needs
What it takes to run GLM 5.3 Flash locally, including quantization sizes, MoE architecture, and hardware like the new Mac Studio and mini PCs.
How to Run Qwen3.8-Flash-Next Locally with llama.cpp
A practical guide to downloading, quantizing, and serving Qwen3.8-Flash-Next locally with llama.cpp, covering VRAM needs on an H100 and quad 3090 rig.
Splitting a 122B MoE Model Across an Nvidia and AMD GPU with Vulkan
How to run a single 122B-parameter MoE model split across mismatched Nvidia and AMD GPUs using Vulkan and llama.cpp for local inference.
Build a Free AI Business Dashboard with ChatGPT Codex
How to build a personal AI dashboard with ChatGPT Codex, combining news, brand mentions, tasks, and newsletters into one control tower app.
How to Price AI Automation Retainers Without Losing Clients
A step-by-step guide to pricing AI automation retainers: diagnose real constraints, pick a KPI, and price against proven value, not guesswork.
How OpenAI and Anthropic Turn Compute Into Profit
A look at the revenue-per-megawatt math behind OpenAI and Anthropic's 2026 shift from venture-funded losses to real profitability.
How to Fine-Tune Qwen3 27B Locally: LoRA, QLoRA and GGUF Guide
Learn how to fine-tune Qwen3 27B on a single GPU using Unsloth, LoRA/QLoRA, and export to GGUF, with dataset creation steps included.
IBM Granite 4.2 3B vs 8B: Local Reasoning Model Tested
Hands-on test of IBM's Granite 4.2 3B and 8B models locally, checking VRAM use, tool calling, and reasoning on real prompts.
How to Run IBM Granite 4.2 Locally with vLLM
Download and serve IBM's Granite 4.2 models locally with vLLM. VRAM needs and setup steps for the 3B, 8B and 30B reasoning variants.
Legora: How a YC-Rejected Legal AI Startup Hit $100M ARR
Legora went from a Y Combinator rejection to $100M ARR in two years. Here's the funding, culture, and product story behind the legal AI startup.
How Much VRAM Do You Actually Need for Local AI in 2026?
How much VRAM local AI actually needs in 2026, from 24GB cards to 512GB Mac Studios, and why quantization changes the math on every build.
Mac Studio M5 Ultra vs DGX Spark: Which Wins for Local AI?
Comparing the upcoming Mac Studio M5 Ultra 512GB to Nvidia's DGX Spark on bandwidth, prefill speed, and price for local AI inference.
How OpenAI Used Codex to Bypass Nvidia's CUDA Moat
OpenAI reportedly used its Codex model to hand-write chip kernels for its new Jalapeno chip, sidestepping Nvidia's decades-old CUDA advantage.
OpenAI's Jalapeno Chip: How It Stacks Up Against Nvidia Blackwell
OpenAI's first custom AI chip, Jalapeno, reportedly beats Nvidia Blackwell on performance-per-watt for LLM inference, per SemiAnalysis lab testing.
OX Alpha on OpenRouter: Free Access, Limits, and How Long It Lasts
OX Alpha is a free stealth model on OpenRouter with a 1M context window and huge rate limits. Here's what's known about access before it disappears.
Unlazy: The Skill That Stops AI Coding Agents From Faking "Done"
Unlazy is a free Claude Code and Codex CLI skill that forces agents to prove work is finished with runnable checks, not confident claims.
What Is Agent-to-Agent Commerce? Inside Stripe's AI Payment Push
Stripe is building machine payment protocols, agent wallets, and stablecoin rails so AI agents can transact directly. Here's how it works.
What Is EVE Preview 4.5B? Tencent's OCR-Free Document Retriever
Tencent's EVE Preview 4.5B retrieves document pages as images using ColBERT-style late interaction, skipping OCR entirely for tables and charts.
DeepSeek V4 Flash on One RTX 3090: Real Tokens-Per-Second Numbers
Real benchmark results for running DeepSeek V4 Flash and Qwen 3.8 27B locally on a single RTX 3090 or 4090 using FreeToken's desktop app.