Claude Code benchmark shows build-first defaults

// 45d agoBENCHMARK RESULT

Claude Code benchmark shows build-first defaults

Amplifying ran Claude Code across 2,430 open-ended prompts in real repos and found a strong build-first bias: custom/DIY beat any single tool in many categories. When it does pick a stack, the defaults skew heavily toward GitHub Actions, Stripe, shadcn/ui, Vercel, PostgreSQL, and Zustand.

// ANALYSIS

This reads less like a product comparison and more like a map of what Claude Code considers “standard” software. For vendors, that’s the uncomfortable part: the agent’s defaults can shape what gets shipped before a human even looks at alternatives.

–Custom/DIY wins in 12 of 20 categories, especially feature flags, auth, caching, and observability, which means the model often prefers to synthesize a solution instead of naming a vendor
–The strongest defaults cluster around the modern JS app stack, with GitHub Actions, Vercel, PostgreSQL, shadcn/ui, Stripe, and Zustand repeatedly surfacing as the path of least resistance
–The study is about revealed preference, not quality or market share, so “rarely picked” does not mean “bad” or “unpopular”
–The practical takeaway is distribution power: if an AI agent is choosing the stack, documentation, examples, and training prevalence matter almost as much as product features

// TAGS

claude-codebenchmarkai-codingagentcliresearch

DISCOVERED

45d ago

2026-04-29

PUBLISHED

45d ago

2026-04-29

RELEVANCE

9/ 10

AUTHOR

Theo - t3․gg

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

OPEN SOURCE37m ago

Vercel Labs' open-source AI browser automation tool, agent-browser, has reached one million weekly downloads on npm, with developer Chris Tate sharing insights on optimizing startup performance.

Chris Tate announced that Vercel Labs' agent-browser, an open-source headless browser automation tool tailored for AI agents, has reached 1,000,000 weekly downloads on npm. Tate noted that a temporary download dip lined up with a transition from running via `npx` to a global install command (`npm i -g`). Implementing a global installation path reduced the tool's startup time to approximately 1 millisecond, which is crucial for low-latency agentic workflows.

MODEL49m ago

Anthropic has launched Claude Fable 5, its latest flagship AI model featuring a 1 million token context window designed for complex agentic workflows and software engineering tasks.

Anthropic has introduced Claude Fable 5, a new flagship AI model optimized for long-running, goal-directed agentic workflows and software engineering tasks. The model boasts a 1 million token context window and is equipped with hard guardrails that fallback to Opus 4.8 for sensitive queries, although access has recently been suspended globally due to U.S. government export control directives.

OPEN SOURCE1h ago

CUA-Gym automates the generation of verifiable reinforcement learning training environments and datasets for computer-use agents to bypass the training data bottleneck.

CUA-Gym is an open-source pipeline and dataset developed by XLang AI to scale Reinforcement Learning with Verifiable Rewards (RLVR) for Computer-Use Agents (CUAs). It addresses the training data bottleneck by co-generating task instructions, environment states, and Python reward functions using a multi-agent Generator-Discriminator-Orchestrator architecture. The framework is paired with CUA-Gym-Hub, which hosts 98 high-fidelity, self-contained mock web applications designed with state injection and session isolation. This setup allows parallel RL training sessions on shared backends without resource interference, producing over 32,000 verified RLVR training tuples that significantly improve CUA performance on benchmarks like OSWorld and WebArena.