SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World
Back to feed
Programming

Developer finds Claude effort setting, not model choice, drove most of a week's quota burn

0
·5 views

A developer exhausted a full week of Claude Ultra quota in three days and initially blamed the model. Benchmarking Opus across effort levels on a routing task showed that costs rose sharply past medium effort, with no gain in correctness. Testing the cheapest model, Haiku, on a regex-engine task matched Python's re on 3,000 patterns, and routine well-specified tasks cost far less on low effort. The author concludes that low effort is safe when outputs can be verified by automated checks, though findings are based on small samples and some runs were excluded. The author also released a free, open-source Claude Code skill that selects models and effort per step and runs only the tests affected by a change.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Open-source helper adds grid-aware accessibility checks for AG Grid in Jest and Playwright

A developer has built ag-grid-a11y-test, a small helper that adds grid-specific accessibility checks on top of general tools like axe-core. The checks cover issues such as row and column counts under virtualization, unnamed icon-only controls, sort state, color-only status indicators, and keyboard arrow navigation. Findings are graded as error, warning, or info, and the rules are drawn from the ARIA grid pattern and typical AG Grid markup. The package is an early alpha not yet published to npm, so it is installed from a GitHub repository, and the author has not yet validated it against every AG Grid version.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Object storage backup costs hinge on retention and deletion controls, not storage rates

A backup alert often reveals a retention design flaw that predates the incident, such as a tenant's stated retention period not matching what the object inventory actually holds. The article argues that comparing scalable object storage for app backups should account for the full lifecycle of data, including recovery points, storage-class transitions, restore reads, and verified deletion, rather than only the headline per-gigabyte price. It recommends tracking logical bytes, changed bytes, recovery-point counts, export reads, and deletion backlog, and alerting on forecast capacity breaches or approaching deletion deadlights instead of simple size thresholds. The author stresses building a workload sheet with consistent units before evaluating vendors such as AWS S3, Azure Blob Storage, and Google Cloud Storage.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Detecting Missed Realtime Messages in Workspace Presence by Refetching Authoritative State

A developer article argues that a connected socket does not prove a displayed participant roster is current, so clients should track the last accepted sequence number. When a gap appears, the client should replace its view with authoritative state rather than reconstruct it from partial events. The author uses a hypothetical auction workspace scenario, with Go as the illustrative language, to show how a client can distinguish a repaired roster from one that only resumed receiving traffic. The piece stresses that sequence numbers need a clear application contract, including a single ordered domain and snapshots that state the sequence they cover. It also recommends refetching after reconnects and tracking unresolved recoveries as a visible failure condition rather than relying on gap frequency alone.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Court Documents Reveal Meta Staff Discussed Discrediting Jonathan Haidt

Court documents reported by The Guardian indicate that Meta employees discussed a "whisper strategy" to undermine social psychologist Jonathan Haidt, author of The Anxious Generation. Staff reportedly weighed ways to get the academic community to disavow him. Haidt's book has helped shape mainstream policy debates about children's online safety. The original article argues that most screen-time proposals focus only on reducing use, overlooking what the screen itself does for users. It also describes Amblyotube, a Meta Quest app that uses independently shaped views for each eye to target vision problems like lazy eye.

0 comments Read more at DEV Community