News & guides
Pinning Model Versions: Why Aliases Break Production Quietly
A stable model alias can point at different weights over time. How that breaks tuned prompts without any error, and what to pin instead.
· 8 min read
Qwen 3.5 Coder vs DeepSeek V4 Pro: Specialist or Generalist
A code-tuned model against a large general MoE. What each design buys you, where the published numbers stop helping, and how to decide on your own repo.
· 9 min read
Qwen 3.6 27B vs Max: Shipped Model or Preview Tier
One is a dense 27B you can run on a single GPU today. The other is a preview whose numbers can move. How to pick a tier inside one model family.
· 9 min readQwen 3.6 Guide: The Frontier Model That Fits on One GPU
Qwen 3.6 27B is dense, not mixture-of-experts, and runs on a single GPU while scoring 77.2 percent. Why that combination matters more than its position on a leaderboard.
· 8 min readQwen 3.6 Max: How to Plan Around a Preview Model
Max is still a preview tier, which means its specs, weights and price can all move. A procedure for evaluating and depending on a model that has not settled.
· 9 min readQwen 3.6 vs MiniMax M3: Different Benchmarks, Really
One reports 77.2 on SWE-bench Verified, the other 59.0 on SWE-bench Pro. Those numbers cannot be subtracted, and the real difference is elsewhere.
· 9 min read
Qwen 3.6 vs Qwen 3.5 Coder: General or Specialised
One is a newer general model, the other a code-specialised predecessor. Whether a specialist still beats a stronger generalist at its own job.
· 8 min read
Regression Testing a Model Upgrade Before It Bites
A new model version is a dependency bump with no changelog. How to gate an upgrade, what to diff, and how to roll back when behaviour shifts.
· 9 min readRunning Frontier Models on Rented GPUs: What It Really Takes
Renting GPUs to serve open weights sounds like a shortcut past API pricing. The memory maths, the utilisation problem, and when it actually pays off.
· 10 min read
Security Review for Open-Weight Models Before You Deploy
Downloading weights means running someone else's artefact inside your network. What to check on provenance, licence, serving stack and data flow.
· 9 min read
SWE-bench Explained: What the Percentage Actually Measures
SWE-bench scores get quoted constantly and understood rarely. What the benchmark does, why its variants are not comparable, and what it cannot tell you.
· 9 min readSWE-bench Leaders Compared: Reading the Scores Correctly
Verified and Pro are different benchmarks and the scores are not comparable. Who leads each variant in 2026, and what the numbers do and do not tell you.
· 9 min readReady to try it?
Every model behind one key. Start free and upgrade when it sticks.