AI Engineering Toolkit
Four tools I built for myself, opened up for every AI engineer and developer. No accounts, no tracking inside the tools, no data leaves your browser.
Paste a prompt, pick a model, see input + output tokens and a per-call, per-1000, and per-million cost across 14 frontier and value models.
Sortable table of context window, max output, modality, input/output pricing, and tier across providers. Find your best $/token tradeoff.
Plan a retrieval system end to end: chunks, vector DB size (float32), one-time embedding cost, and monthly cost across Pinecone, pgvector, and Chroma.
Word-level diff between two prompts with token counts on each side and a delta. Spot prompt drift across iterations.
Why this exists
No signup
Most calculators want your email. These four run entirely in your browser — no API keys, no accounts, no telemetry.
Pricing that is current
Provider pricing changes monthly. This data is refreshed for late 2025 and called out wherever it is approximate.
Built by an engineer using them
I use these every day to scope client RAG work and ship features at PaisaReality and Epicenter Exchange. If a number lies, I notice fast.