Comments URL: https://news.ycombinator.com/item?id=47844851

Points: 14

# Comments: 3

But OpenAI might be running a playbook most competitors can’t afford to match: - Lock distribution now - Trade margins for data + ubiquity - Introduce upsells / infra plays later

With ₹399 and no credit card barrier, this could be AI's “Jio moment.” Suddenly, millions of first-time payers can access GPT-5.

Imagine the ripple effects on education, vernacular content, coding bootcamps, and small-town creators.

Yes, you can downgrade from the $20 Plus plan to ChatGPT Go. The switch takes effect at the end of the current billing cycle.

₹399/month (roughly $4.80), with the support of UPI payments. And this wasn't a loud rollout at all, just a quiet addition to their pricing page. Now the point is that this isn't just a discounted GPT-Plus plan but it's actually built for India's mass market. Think of the students and everyday users living beyond tier 1 cities, who've never paid for AI before but know what ChatGPT is. The highlight is that along with credit cards they have added the support of UPI which is widely accepted across the country because let's face it that not everyone owns a credit card in India.

What’s included:

- GPT-5 (with extended usage)

- Image generation

- File uploads

- Python tools, memory, custom GPTs

What’s missing:

- GPT-4o or API access

- Connectors, Sora, or enterprise features

- No annual billing or bundles

They’re clearly not targeting the English-speaking dev crowd that already uses ChatGPT. This feels more like a test run for mass-market localization at scale, with India as the first sandbox.

$20/month doesn’t work in a country where Netflix costs ₹149. But ₹399 with UPI, is an unlock.

Feels like OpenAI is prepping for the next 500 million users, not the next 500 YC-backed teams.

Docs: https://help.openai.com/en/articles/11989085-what-is-chatgpt...

Would love to hear from anyone testing the usage caps.

Comments URL: https://news.ycombinator.com/item?id=44948473

Points: 11

# Comments: 17

Retroactive adjustments are handled by versioning usage records, so instead of overwriting historical events, we apply corrections as delta events. Invoices pull from the latest state, but we preserve full audit logs underneath to avoid integrity gaps.

I’ll DM you, we can run a quick POC and see if it fits your setup. Appreciate the interest.

This week, we’re doing a 5-day launch week, where we’re shipping a new set of billing features every day. Github link: https://github.com/flexprice/flexprice

We’re building Flexprice specifically to avoid that constant rewrite cycle, good to know it resonates. Happy to chat if you run into edge cases while scaling those new plans.

Good luck with your GenAI launch!

For internal retries, we batch in-memory and attach unique IDs before dispatch to avoid double-counting.

Our approach focuses on: - Fire-and-forget ingestion with in-memory queues so events don’t block product requests - Strict idempotency tokens tied to every event, enforced at the API layer - Lightweight retry logic that prevents double-counting but guarantees delivery under transient failures

Storage-wise, we’ve leaned on a mix of time-series DBs for raw events and pre-aggregated summaries for billing views.

Would love to swap notes on failure patterns or queue setups if you’ve dealt with similar scale.

Flexprice is an open source monetization platform for AI & Agentic companies. You can build usage-based, credit-based, or hybrid pricing models with full control.

Usage-based billing works on paper, but breaks fast in production:

(1) High-volume data (e.g., 10k API calls/sec from AI inferences) breaks naive pipelines, dropping events and causing underbilling or customer complaints.

(2) Backfilling missing events or retroactive discounts screws up invoices, forcing 3 AM log stitching to avoid revenue leaks.

(3) Prepaid credits and trials are a mess because it requires you to handle expirations, partial deductions, and priority rules.

(4) Product teams want real-time, per-customer usage reports (e.g., “How many tokens did user X consume?”), but querying raw events requires slow, custom SQL hacks.

(5) Webhook integrations for invoicing or CRMs fail on duplicates, out-of-order events, or time zone bugs, risking customer disputes.

(6) What works for 100 users fails at 10,000 because rate limits, and idempotency leads to urgent rewrites.

This week, we’re doing a 5-day launch week, where we’re shipping a new set of billing features every day. Today’s launch includes:

(1) Fire-and-forget event ingestion with SDKs (Python, JS, Go) and auto retry handling

(2) Real-time event debugger for tracking ingestion status, filtering payloads, and inspecting failures

(3) Usage analytics API providing billing-grade summaries with filters, groupings, and time-bucketed results

(4) Prepaid credit system supporting trial credits, expirations, and configurable priority on deduction

(5) Webhook system to automate billing workflows like invoicing or subscription changes

We’re shipping new features daily. It’s early, actively evolving, and contributions are welcome. If you’ve dealt with usage-based billing, broken credit logic, or event ingestion at scale, your feedback would help.

You can read the details of the launch at https://flexprice.io/blog/the-developer-toolkit. You can clone the project and explore the repo at https://github.com/flexprice, and demo at https://www.youtube.com/watch?v=SFARthC7JXc

Would love to hear your technical critiques, edge case concerns, or how you’ve handled billing infrastructure in your own stack.

Comments URL: https://news.ycombinator.com/item?id=44422663

Points: 7

# Comments: 14

This guide includes: - Configuring recurring credit grants (e.g., 100 credits/month) - Capping rollover at 2× monthly allocation - Real-time metering (e.g., 10 credits to create a table; 1.5 per-row on enrichment) - Monthly vs annual billing models, credit expiry rules

This addresses a challenge many SaaS/AI/API products face: building transparent, usage-aligned pricing that’s easy to iterate on.

Would be grateful for HN feedback especially around edge cases or UI/UX when exposing credit consumption to users.

Comments URL: https://news.ycombinator.com/item?id=44245780

Points: 2

# Comments: 3

Curious about your approach to the networking stack. Are you planning to support more protocols like HTTPS or WebSockets in the future, or is the focus more on keeping this lightweight and minimal for now?