1 firsthand and 1 secondhand accounts in the October guide. The sample is selected and not representative.
Firsthand account · Simon Willison · 2026-10-09
A Django feature shipped after a second round of review
Codex voice mode · GPT-6 Astra High
Willison reports building newsletter models, migrations, imports, archive pages and search integration with Codex voice mode, using GPT-6 Astra High. He then reviewed the pull request, changed a subprocess-based import to an API-based import, and refined the display before deploying.
What this does not establish: One experienced developer and one Django feature. He describes roughly half an hour by voice plus further typed work; this is not a timing benchmark or a general result for every Codex model.
An older destructive-action incident shows why agent permissions matter
Codex with full-access execution · GPT-5.6
Willison collected an OpenAI engineering statement describing reports of unintended file deletion involving GPT-5.6, full-access execution and temporary-directory handling. It illustrates a failure mode worth isolating when an agent can modify your files.
What this does not establish: A July report about an older model, relaying an engineering statement. It does not establish a current failure rate, an unresolved current bug, or that another provider is immune.
Terminal coding workflows are included with Claude Pro and both Max tiers. The same subscription also covers Claude chat.
How access is sold: Pro is $20/month; Max 5x is $100 and Max 20x is $200. Five-hour and separate weekly limits apply; multipliers describe allowance, not completed features.
2 firsthand and 0 secondhand accounts in the October guide. The sample is selected and not representative.
Firsthand account · Simon Willison · 2026-09-07
A browser video tool built for a specific practical need
Claude Code for web · Claude Fable 5.1
Willison describes asking Claude Fable 5.1 in Claude Code for web to build a browser video compressor using FFmpeg compiled to WebAssembly. The linked tool lets users compare encoded outputs and choose a file size/quality trade-off.
What this does not establish: A focused browser utility, not a production SaaS or a comparison against Codex. The report concerns that configured model and date, not every current Claude release.
A growing codebase needed context discipline and two review gates
Claude Code
Nnorukam’s account of maintaining a trading application with Claude Code emphasizes bounded context, task plans, automated checks, a separate manual review and shadow deployment. He also describes costly unverified output and repeated context rebuilding before changing that workflow.
What this does not establish: Author-reported production experience, updated September 18. The author sells Claude-related kits and discloses affiliate links. We have not independently verified the reported error rates, uptime, code size or trading results, and do not use them as quality scores.
Listed currency: USD · Monthly billing. Region is unverified unless specified per offer. Annual discounts are excluded. These are published amounts, not a local checkout quote; confirm region, taxes and availability with the provider. Subscription allowance and API token billing are different quantities.
5x the Pro per-session usage allowance; session resets every five hours; a separate weekly limit applies across models. Includes Claude Code; not 5x completed features. Web subscription price; mobile pricing may differ. Limit/access source ↗
20x the Pro per-session usage allowance; session resets every five hours; a separate weekly limit applies across models. Includes Claude Code; not 20x completed features. Web subscription price; mobile pricing may differ. Limit/access source ↗