Model comparisons, migration notes, and essays on agency and measurement. Browse Models & tooling for model and CLI decisions, or Open questions for essays on human judgment. For scheduling, permissions, and troubleshooting, visit Guides.
-
GPT-6 vs GPT-6.1: Astra reasoning effort and model selection
Compare GPT-6 Astra effort, Max and Ultra, GPT-6 Sol and 6.1 Sol, and starting settings for coding, writing, research, and repeatable work.
-
Claude Opus 5.5 vs. Sonnet 5.5: which model fits your work?
Compare Claude Opus 5.5 and Sonnet 5.5 using official benchmarks and API prices, then choose a model and effort level for coding, documents, research, and agents.
-
Qualify an agent CLI for unattended use in four probes
Before an agent CLI runs on a schedule, four cheap experiments establish what its flags actually do: whether restrictions hold, what refusal looks like from the outside, how…
-
Reasoning effort changed what my agent tried, not how well it did
Twelve timed runs of one arithmetic task across four effort levels. Latency did not separate the levels and neither did correctness. What separated them was behavior: the two…
-
When an AI rule changes, who gets the correction?
A four-way change test and five continuity duties for deciding when an AI correction should reach people affected by an earlier version.
-
When does a local AI answer measure the model?
Use a four-layer test to tell whether a local AI failure belongs to the model, runtime, application, or host.
-
Qwen 3.8 27B thinks by default, and the default effort is xhigh
See the official Qwen 3.8 27B defaults, then change effort or turn analysis off before judging speed or quality.
-
Private AI can protect your input without making every outcome private
See what encrypted inference protects, which privacy layers remain separate, and how to assess a private-AI claim without overstating it.
-
Who Spends the Time Your AI Saves?
A four-condition test for when an AI agent saves its user’s time by creating review and response work for other people.
-
Who Is Free-Riding on AGI?
A concise argument that the free-rider label should be tested against three ledgers: development cost, wider contribution, and control of the result.
-
Did I Mean It Before the Machine Said It?
A philosophical argument about what happens when AI-generated language arrives before settled intention, is attributed backward as an original thought, and is enacted forward as a commitment.
-
GitHub Copilot Retired Gemini 2.5 Pro and Gemini 3 Flash: What to Switch To
The July 31 replacements have since retired: Gemini 3.1 Pro on September 1 and Gemini 3.6 Flash on October 2, 2026. Check GitHub’s current supported-model list, plan, client,…