News & guides
Vendor-Claimed vs Independent Benchmarks: Reading Both
A launch-post score and a third-party score measure different things. Where the gap comes from, and how to normalise figures before you compare them.
· 9 min read
Vercel AI SDK: Wiring Up a Custom OpenAI-Compatible Provider
Point the AI SDK at your own base URL, stream to a React client, keep tool calls working, and avoid the edge runtime traps that break streaming in production.
· 9 min read
Verifier Agents: Gating Completion on a Check That Passes
The most reliable agent improvement is refusing to accept completion until something verifies it. How to build a verifier that is worth trusting.
· 9 min read
Vision-Language Models: What to Test Before You Commit
Native image input is now common in open-weight models, but published multimodal benchmarks predict almost nothing. Three probes that separate them.
· 9 min read
Vocabulary Size Trade-offs in Language Model Tokenizers
A bigger vocabulary shortens sequences and costs parameters and rare-token quality. What the dial actually controls and how it reaches your bill.
· 9 min read
VS Code LLM Extension Setup: Custom Endpoints That Work
A provider-agnostic guide to wiring any VS Code AI extension to your own endpoint — where settings live, what the extension cannot infer, and how to debug it.
· 8 min read
What a Model Upgrade Costs Beyond the Per-Token Rate
A newer model at a lower rate can still raise your monthly bill. Regression testing, prompt drift, output length and tool-call changes are where the cost actually lands.
· 9 min read
When to Switch Models: A Threshold Worth Crossing
New models ship constantly and switching is never free. The signals that justify a migration, the ones that do not, and how to run the decision.
· 9 min read
Where Quantised Models Actually Lose Quality
Quantisation degrades capabilities unevenly rather than uniformly. Which ones erode first, why a benchmark delta hides it, and how to measure the loss on your own work.
· 9 min read
Which Open Model Fits Your GPU: A Memory-First Guide
Total parameters set the memory floor, active parameters set the speed. How to work out what your hardware can actually serve before reading a benchmark.
· 9 min read
Why Benchmarks Disagree, and Which One to Believe
Two credible evaluations can rank the same models differently without either being wrong. How to read that disagreement instead of picking the flattering one.
· 9 min readWindsurf Setup Guide: Custom Models and the BYOK Question
What Windsurf needs to talk to a model you chose, how to find the provider settings in your build, and the local proxy fallback when there is no slot for one.
· 8 min readReady to try it?
Every model behind one key. Start free and upgrade when it sticks.