Two important model launches and a practical guide to keeping the database standing as the product grows.
Claude Opus 5 gets close to Fable 5 at half the price
Anthropic released Claude Opus 5, which according to the company comes close to Claude Fable 5 at half the price and costs the same as Opus 4.8. It leads coding and knowledge work evaluations such as Frontier-Bench and GDPval-AA and, on Zapier AutomationBench, which measures end-to-end business tasks, passes about 1.5 times as many tasks as the runner-up at the same cost. The model also verifies its own work better and iterates until it succeeds.
Automating whole business processes, not just isolated steps, becomes more reliable. Revisit flows that used to need heavy human supervision.
How Dexpert helps: Artificial IntelligenceSources
- Anthropic: Introducing Claude Opus 5
Gemini 3.6 Flash uses 17% fewer tokens and 3.5 Flash-Lite reaches 350 tokens per second
Google released Gemini 3.6 Flash, with better coding, knowledge work and multimodal performance and, according to Artificial Analysis, 17% fewer output tokens than 3.5 Flash, at a lower cost per token. It also launched 3.5 Flash-Lite, the fastest and cheapest model of the 3.5 generation, at 350 output tokens per second. Google says 3.5 Pro is testing with partners and that pre-training of Gemini 4 has started.
High-volume agents, such as customer service and document classification, get cheaper and faster. Using Flash-Lite for simple steps and a bigger model only where needed cuts the total cost.
How Dexpert helps: Artificial Intelligence
Hatchet gathers two years of lessons to keep Postgres up as a startup grows
Hatchet published the internal guide it wrote for its engineers from two years of running Postgres in production. It goes from basics, such as designing the schema around the most common queries and using compound indexes aligned with ORDER BY, to topics such as the query planner, bulk updates, autovacuum tuning, whose default settings can take the database down, and FOR UPDATE SKIP LOCKED. The author notes that many optimizations require stepping outside the ORM and writing SQL.
Database problems tend to show up exactly when the business grows. A review of schema, indexes and configuration before the peak avoids slowdowns and downtime.
How Dexpert helps: Backend & APIsSources
Want to put one of these to work in your company?
Dexpert builds custom systems, apps, integrations and AI solutions. Tell us what you need and get an estimate, or take the free digital maturity assessment.