# Cafiyn Pulse

> Cafiyn Pulse is a free suite of three pre-launch tools for founders, indie hackers, vibe coders, solo builders, and small engineering teams: a stack-readiness grader, an AI cost comparator across 79 models from 22 providers, and a stack-tailored launch checklist. No account is required to use any of them. Built by Cafiyn Innovations (Bangalore, India).

## Provenance

Model prices are read from each provider's own pricing page. An automated drift check runs weekly and flags any published price that no longer appears at its source; a person re-verifies the full roster monthly. Roster version 2026-09-05, model pages verified 2026-09-05, last full reconciliation 2026-09-07. Every change, including our own corrections, is dated and sourced in the change log below.

## Tools

- [Stack Grader](https://pulse.cafiyn.com/stack-grader): 3-minute self-audit across 10 production categories (auth, secrets, backups, rate limiting, payments, monitoring, HTTPS, validation, access control, dependencies). Returns an A-F grade plus a prioritised fix plan with named tools.
- [Cost Comparator](https://pulse.cafiyn.com/cost-comparator): AI cost calculator across 79 models. Current text models: Anthropic (Claude Fable 5.1, Claude Opus 5, Claude Sonnet 5, Claude Haiku 4.5); OpenAI (GPT-6 Astra, GPT-5.6 Sol, GPT-5.6 Terra, GPT-5.6 Luna, GPT-5.5, GPT-5.4, GPT-5.4 mini, GPT-5.4 nano, GPT-5, GPT-5 mini, GPT-5 nano, o3 (reasoning), o3-mini (reasoning), o4-mini (reasoning), GPT-4o, GPT-4o mini); Google (Gemini 3.8 Flash, Gemini 3.7 Flash, Gemini 3.5 Flash, Gemini 3.5 Flash Lite, Gemini 3.1 Pro); DeepSeek (DeepSeek V4 Pro, DeepSeek V4 Flash); xAI (Grok 4.6, Grok 4.3); Mistral (Mistral Large); Together (Llama 3.3 70B (Together)); Groq (GPT OSS 120B (Groq), GPT OSS 20B (Groq)); Perplexity (Sonar (web-aware), Sonar Pro (web-aware), Sonar Reasoning Pro). Also voice, image and video. Live math on your own usage numbers.
- [Launch Checklist](https://pulse.cafiyn.com/launch-checklist): Stack-tailored pre-launch checklist for 11 stacks. Every item has why, what and how, with optional notes and proof URLs.

## AI model pricing

Per-model pricing pages, each showing cost per million input and output tokens, a sample-request cost, and a head-to-head table. Verified 2026-09-05.

- [AI model pricing (hub)](https://pulse.cafiyn.com/cost-comparator/pricing): all 12 models in one table.
- [LLM API price changes, tracked](https://pulse.cafiyn.com/cost-comparator/pricing/changes): dated, sourced record of every price change we verify, including increases already scheduled and our own published corrections. This is original data.
- [Claude Sonnet 5 pricing](https://pulse.cafiyn.com/cost-comparator/pricing/claude-sonnet-5) ($2.00 input, $10.00 output per million tokens)
- [Claude Opus 5 pricing](https://pulse.cafiyn.com/cost-comparator/pricing/claude-opus-5) ($5.00 input, $25.00 output per million tokens)
- [Claude Haiku 4.5 pricing](https://pulse.cafiyn.com/cost-comparator/pricing/claude-haiku-4-5) ($1.00 input, $5.00 output per million tokens)
- [GPT-6 Astra pricing](https://pulse.cafiyn.com/cost-comparator/pricing/gpt-6-astra) ($10.00 input, $50.00 output per million tokens)
- [GPT-5 pricing](https://pulse.cafiyn.com/cost-comparator/pricing/gpt-5) ($1.25 input, $10.00 output per million tokens)
- [GPT-5.6 Sol pricing](https://pulse.cafiyn.com/cost-comparator/pricing/gpt-5-6-sol) ($4.00 input, $20.00 output per million tokens)
- [Gemini 3.1 Pro pricing](https://pulse.cafiyn.com/cost-comparator/pricing/gemini-3-1-pro) ($2.00 input, $12.00 output per million tokens)
- [Gemini 3.8 Flash pricing](https://pulse.cafiyn.com/cost-comparator/pricing/gemini-3-8-flash) ($0.75 input, $3.75 output per million tokens)
- [DeepSeek V4 Pro pricing](https://pulse.cafiyn.com/cost-comparator/pricing/deepseek-v4) ($0.66 input, $1.98 output per million tokens)
- [Grok 4.6 pricing](https://pulse.cafiyn.com/cost-comparator/pricing/grok-4-6) ($2.00 input, $6.00 output per million tokens)
- [Llama 3.3 70B pricing](https://pulse.cafiyn.com/cost-comparator/pricing/llama-3-3-70b) ($1.04 input, $1.04 output per million tokens)
- [Mistral Large pricing](https://pulse.cafiyn.com/cost-comparator/pricing/mistral-large) ($0.50 input, $1.50 output per million tokens)

## Cost Comparator by workload

- [Chatbot / support assistant](https://pulse.cafiyn.com/cost-comparator/chatbot): Support bots and product assistants are the most common AI workload, and the easiest to under-budget. Here is the real math.
- [Voice agent](https://pulse.cafiyn.com/cost-comparator/voice-agent): Voice is three billed legs per turn: transcription, the model, and speech synthesis. Most estimates only price one of them.
- [Image generation](https://pulse.cafiyn.com/cost-comparator/image-gen): Per-image pricing looks trivial until you multiply it by users who generate more than once.
- [RAG / knowledge assistant](https://pulse.cafiyn.com/cost-comparator/rag): Retrieved context is the hidden cost driver in RAG. It is paid as input tokens on every single query.
- [Multi-modal app](https://pulse.cafiyn.com/cost-comparator/multi-modal): Text, voice, and image each have their own pricing unit. A multi-modal product pays all of them per user, per day.
- [Code assistant](https://pulse.cafiyn.com/cost-comparator/code-assistant): Coding workloads send more tokens per call than almost any other use case. Context window, not call count, drives the bill.

## Launch Checklist by stack

- [Lovable launch checklist](https://pulse.cafiyn.com/launch-checklist/lovable): Lovable ships a working app fast. "Working" and "production-ready" are not the same checklist.
- [Bolt.new launch checklist](https://pulse.cafiyn.com/launch-checklist/bolt): Bolt.new is built for speed of iteration. The checklist below is the speed-of-iteration tax you pay back before real users show up.
- [Replit Agent launch checklist](https://pulse.cafiyn.com/launch-checklist/replit): Replit hosting makes shipping trivially easy. Easy hosting is not the same as production hardening.
- [v0 (Vercel) launch checklist](https://pulse.cafiyn.com/launch-checklist/v0): v0 is excellent at UI. Production readiness lives in the layers v0 was never asked to think about.
- [Next.js (custom) launch checklist](https://pulse.cafiyn.com/launch-checklist/nextjs): Hand-rolled means nothing is generated for you, including the parts people forget on purpose-built apps too.
- [React Native / Expo launch checklist](https://pulse.cafiyn.com/launch-checklist/react-native): Mobile has a review gate a website does not. Miss one of these and you lose a week to a rejection, not a bug report.
- [Flutter launch checklist](https://pulse.cafiyn.com/launch-checklist/flutter): One codebase, two review processes with different rules. Both need a separate pass.
- [WordPress launch checklist](https://pulse.cafiyn.com/launch-checklist/wordpress): The most-targeted CMS on the internet needs the most deliberate security pass, not the least.
- [Shopify launch checklist](https://pulse.cafiyn.com/launch-checklist/shopify): Shopify handles the hard infrastructure. Your checklist is everything layered on top of it.
- [Webflow launch checklist](https://pulse.cafiyn.com/launch-checklist/webflow): Webflow removes the build step. It does not remove the need to check your own configuration.
- [Custom / other stack launch checklist](https://pulse.cafiyn.com/launch-checklist/custom): The categories that matter do not change based on your framework. The specific implementation does.

## Stack comparisons

- [Supabase vs Firebase](https://pulse.cafiyn.com/vs/supabase-vs-firebase): Database / BaaS. Supabase and Firebase both give you auth, database, storage, and functions with one signup. Pick Supabase if you want real Postgres, SQL, and open-source portability. Pick Firebase if you are shipping a mobile-first app or want Google-scale infra with the least backend code.
- [Supabase vs Neon](https://pulse.cafiyn.com/vs/supabase-vs-neon): Postgres BaaS. Both give you managed Postgres, but they solve different problems. Supabase is a Firebase-style all-in-one backend built on Postgres. Neon is a serverless Postgres engine you plug into your own stack. Pick Supabase if you want one vendor for everything; pick Neon if you want the best Postgres in class and are happy composing auth + storage + functions yourself.
- [Neon vs PlanetScale](https://pulse.cafiyn.com/vs/neon-vs-planetscale): Serverless database. Neon and PlanetScale are both serverless databases but they target different worlds. Neon is Postgres with branching and autoscale, great for AI apps, Next.js, and preview environments. PlanetScale is MySQL with Vitess-scale sharding, great for legacy migrations and huge write workloads. Most AI-era startups want Postgres, so Neon wins the majority.
- [Vercel vs Netlify](https://pulse.cafiyn.com/vs/vercel-vs-netlify): Frontend hosting. Vercel and Netlify both do git-push-to-deploy for static + JS. Vercel wins if you use Next.js; Netlify wins if you use everything else and want form + identity features baked in. For most modern React / SvelteKit / Astro apps, Vercel has more momentum in 2026.
- [Vercel vs Cloudflare Pages](https://pulse.cafiyn.com/vs/vercel-vs-cloudflare-pages): Frontend hosting. Vercel gives you the best Next.js experience with excellent DX. Cloudflare Pages gives you unlimited bandwidth on a much larger edge network and a real backend via Workers. Pick Vercel for speed of shipping Next.js; pick Cloudflare Pages for cost efficiency, global reach, and full-stack on the edge.
- [Render vs Railway](https://pulse.cafiyn.com/vs/render-vs-railway): Full-stack hosting. Render and Railway are the two go-to Heroku replacements in 2026. Render is predictable and cheaper at scale. Railway is faster to spin up and has better preview environments. Both host Node, Python, Go, Postgres, Redis, and static sites.
- [Render vs Fly.io](https://pulse.cafiyn.com/vs/render-vs-fly-io): Global infrastructure. Render is the Heroku for the modern era: predictable pricing, managed everything, zero-ops. Fly.io is the global-edge-anywhere platform: deploy your app in Tokyo, Sydney, Frankfurt at once. Pick Render for simple ops; pick Fly for global latency requirements.
- [Clerk vs Auth0](https://pulse.cafiyn.com/vs/clerk-vs-auth0): Auth. Clerk is the modern React-native auth service with prebuilt UI and org management. Auth0 is the enterprise-standard identity platform. Clerk wins for startups shipping fast; Auth0 wins for teams selling into enterprise from day one.
- [Clerk vs Supabase Auth](https://pulse.cafiyn.com/vs/clerk-vs-supabase-auth): Auth. Clerk is a dedicated auth SaaS with beautiful prebuilt UI and org management. Supabase Auth is built into every Supabase project and integrates cleanly with Postgres row-level security. Pick Clerk if you want auth done in 15 minutes with prebuilt UI; pick Supabase Auth if you already use Supabase and want auth as a native part of your DB.
- [Resend vs SendGrid](https://pulse.cafiyn.com/vs/resend-vs-sendgrid): Transactional email. Resend is the modern developer-first email API, React templates, clean dashboard, quick setup. SendGrid is the battle-tested Twilio-owned incumbent, with more analytics and higher scale but a dated developer experience. For startups shipping in 2026, Resend is the clear pick; for enterprise-scale, SendGrid still wins on capabilities.
- [Lovable vs Bolt.new](https://pulse.cafiyn.com/vs/lovable-vs-bolt): AI app builders. Lovable and Bolt.new are the two most popular AI app builders in 2026. Lovable is optimised for building full apps with Supabase + GitHub. Bolt.new is optimised for lightning-fast prototyping in the browser. Pick Lovable to ship real products; pick Bolt.new for quick prototypes and experiments.
- [Lovable vs v0 (Vercel)](https://pulse.cafiyn.com/vs/lovable-vs-v0): AI app builders. Lovable and v0 are the two flagship AI app builders in 2026. Lovable generates full-stack apps (with Supabase). v0 focuses on UI-heavy Next.js apps with Tailwind + shadcn/ui. Pick Lovable to ship a full product with DB and auth; pick v0 if your project is UI-heavy and you already live in the Vercel + Next.js world.
- [Lovable vs Replit](https://pulse.cafiyn.com/vs/lovable-vs-replit): AI app builders. Lovable and Replit both let you build an app by prompting an AI agent, but they diverge on where the app ends up living. Lovable generates a web app with Supabase and syncs to GitHub, so you own the code and host it anywhere. Replit's Agent builds inside Replit itself, with hosting, a database, and deployment already part of the platform, no separate host to pick.
- [Stripe vs Paddle](https://pulse.cafiyn.com/vs/stripe-vs-paddle): Payments. Stripe and Paddle solve the same problem from opposite directions. Stripe gives you the best payment API and leaves sales tax and VAT compliance to you. Paddle charges more per transaction but becomes the legal seller and handles tax in every country automatically. Pick Stripe if you want full control and are comfortable with tax compliance yourself. Pick Paddle if you are selling globally as a solo founder and do not want to register for VAT in dozens of countries.
- [PostHog vs Mixpanel](https://pulse.cafiyn.com/vs/posthog-vs-mixpanel): Product analytics. PostHog and Mixpanel both answer "what are users actually doing in my product," but PostHog bundles analytics with session replay, feature flags, and A/B testing in one open-source tool, while Mixpanel stays focused and does funnel and retention analysis extremely well. Pick PostHog if you want one tool instead of four. Pick Mixpanel if you want the sharpest possible analytics UI and are fine adding other tools separately.

## Startup Signals, weekly briefing

- [Startup Signals (current issue)](https://pulse.cafiyn.com/signals): issue 2, Week of 7 September 2026. Nvidia confirmed a $12.93B acquisition of Hugging Face, the registry three million models and eighteen million developers already depend on. In the same window OpenAI's deprecation page filled up: seven shutdown dates between 23 October and 26 February, including o3-mini, Agent Builder, the whole audio family and Whisper. One of them is seventeen days away and has no replacement.
- [RSS feed](https://pulse.cafiyn.com/signals/feed.xml) and [JSON Feed](https://pulse.cafiyn.com/signals/feed.json): full text of every issue and analysis page.
- [API changes and deprecations](https://pulse.cafiyn.com/signals/topic/api-changes)
- [Shutdowns and wind-downs](https://pulse.cafiyn.com/signals/topic/shutdowns)
- [Funding and benchmarks](https://pulse.cafiyn.com/signals/topic/funding)
- [AI tooling and models](https://pulse.cafiyn.com/signals/topic/ai-tooling)

### Deep analysis

- [OpenAI has published seven shutdown dates. Here is the calendar.](https://pulse.cafiyn.com/signals/story/openai-deprecation-calendar-2026): Seven dates between 23 October and 26 February. Every one announced months ahead, in public. The failure mode is never the notice.
- [Nvidia bought Hugging Face. Your open-weight fallback is now a vendor relationship.](https://pulse.cafiyn.com/signals/story/nvidia-hugging-face-portability): The commitments say nothing breaks. The structural change is that "we can always self-host" is now a dependency on one vendor rather than on a neutral commons.
- [The Videos API goes away on 24 September and OpenAI names nothing to move to.](https://pulse.cafiyn.com/signals/story/sora-2-videos-api-removal): Every other deprecation on OpenAI's page offers a migration target. This one does not, which makes it a different kind of problem.
- [Whisper shuts down on 26 February. The risk is not the deadline.](https://pulse.cafiyn.com/signals/story/whisper-deprecation-migration): Five months is a generous window. Whisper is also typically wired into a background job that fails into a queue rather than into a user's face.
- [The Assistants API is gone. Your app may not have told you yet.](https://pulse.cafiyn.com/signals/story/openai-assistants-api-sunset): It shut down on 26 August 2026. The replacement is not a drop-in, and the most common migration mistake changes your model's behaviour without throwing an error.
- [One Copilot code review costs 13 requests. Here is the rest of the maths.](https://pulse.cafiyn.com/signals/story/github-copilot-usage-based-billing): Copilot now bills on usage rather than seats. The headline change is simple. The multipliers underneath it are where teams get surprised.
- [Row-level security is on by default now. That is not the same as being protected.](https://pulse.cafiyn.com/signals/story/lovable-rls-default-not-protection): Lovable fixed the default. The gap that actually leaks data was never the default, it was the policy, and no platform can write that for you.
- [Your model comparison is measuring the wrong number.](https://pulse.cafiyn.com/signals/story/cost-per-task-not-cost-per-token): Every price list is per token. Your bill is per finished piece of work. Those two numbers can rank models in opposite orders.
- [An agent with production log access is a new identity. Treat it like one.](https://pulse.cafiyn.com/signals/story/coding-agents-production-access): Always-on agents that read production logs are arriving fast. The access-control question they raise is old and well understood, which is the good news.

## Guides

- [All free tools](https://pulse.cafiyn.com/free)
- [AI cost calculator guide](https://pulse.cafiyn.com/free/ai-cost-calculator)
- [Production readiness checklist](https://pulse.cafiyn.com/free/production-readiness-checklist)
- [Launch checklist guide](https://pulse.cafiyn.com/free/launch-checklist)
- [Learn](https://pulse.cafiyn.com/learn)
- [My Startup, founder guide](https://pulse.cafiyn.com/mystartup)
- [AEO / GEO / SEO](https://pulse.cafiyn.com/aeo-geo-seo)

## About

- [About and methodology](https://pulse.cafiyn.com/about): who publishes this, how the Stack Grader scores, how prices are verified, and the corrections policy.
- Publisher: Cafiyn Innovations, Bangalore, India. Founder and author: Karthik Kumar. Contact: hello@cafiyn.com.