100% valid JSON across OpenAI/Anthropic with automatic model selection

Show HN: 100% valid JSON across OpenAI/Anthropic with automatic model selection

100% valid JSON across OpenAI/Anthropic with automatic model selection is a business in show hn launch.

Profile

Hey HN,I got tired of JSON parsing failures and manually switching between LLMs on task basis. So, I built Perf, a single endpoint that picks the right model and guarantees valid JSON back.The problem: LLMs are unreliable with structured output. You end up writing retry logic, validation, fallback chains or you end up overpaying for a flagship model for trivial tasks.What Perf does: You call perf.chat(messages) and it analyzes the prompt, routes to the best model for cost/quality, validates the output, and repairs or retries if needed.Benchmark results (~500 prompts):- 100% valid JSON (vs ~85-95% from direct API calls)- 90% cheaper than GPT-4o baseline (with identical prompts and no retries)

The company sits in the show hn launch segment, where it competes on its core product or service.

The company is listed at the Show HN launch stage of development.

Website: demo.withperf.pro