Up to $120,000 total across four tiers: $1,000 (Launcher), $20,000 (Gold), $40,000 (Platinum), and $59,000 (Diamond). The value is real because service credits apply to most core products (compute, databases, storage, networking, AI/ML), not just a narrow subset. What you can do depends on what you run, but the practical use is covering pay-as-you-go infrastructure while you build and scale. If you join via the AI Catalyst variant, you also get up to 2 billion Model Studio tokens for Qwen-series models, which can offset a lot of inference testing.

Yes. Account creation requires a payment method.

It depends on the tier and starts from redemption: Launcher lasts 3 months, Gold 6 months, Platinum 9 months, and Diamond 12 months.

Yes. If you have Alibaba Cloud credits you won’t use before they expire, you can list them on AI Credit Mart and sell them at up to 70% of face value. Companies regularly list surplus credits from startup programs and enterprise agreements.

AI Credit Mart has discounted Alibaba Cloud credits available from companies with surplus allocations. Prices are typically 30-70% below retail.

Unused credits expire per tier, and pay-as-you-go services can start billing your payment method at standard rates once the credits are exhausted. Excluded services (like domains, CDN, Marketplace, Academy courses, and prepaid products) are billed normally even before expiration, so you can still see charges. If you are not upgrading to paid plans, back up data first because resources may be deleted if you don’t move to a paid setup.

Yes. It includes the same tiered service credits, plus AI-specific perks like up to 2 billion Model Studio tokens, 1:1 AI office hours, and additional program benefits.

Not completely. The free tokens cover real-time inference only, not batch calls or context cache, and the quota is shared across the Alibaba Cloud account and its RAM sub-users.

Up to $20,000 of Claude API usage over 6 months. Anthropic’s own math says that’s roughly 20 billion input tokens or about 4 billion output tokens at Haiku 4.5 pricing, or roughly 4 billion input tokens or about 800 million output tokens at Opus 4.5 pricing. In practice, that can cover anything from large-scale literature extraction and structured summaries to repeated analysis passes across big text corpora. If you mix models (Haiku for bulk, Sonnet/Opus for the hard parts), the credits usually go much further than teams expect.

No. Anthropic awards credits directly to your organizational API account.

These program credits last for 6 months from award.

Yes. If you have Anthropic credits you won’t use before they expire, you can list them on AI Credit Mart and sell them at up to 70% of face value. Companies regularly list surplus credits from startup programs and enterprise agreements.

AI Credit Mart has discounted Anthropic credits available from companies with surplus allocations. Prices are typically 30-70% below retail.

There’s no auto-billing: when credits run out, API access stops unless you add a payment method.

You get access to Anthropic’s standard model suite via the API: Claude Opus 4.5, Sonnet 4.5, Haiku 4.5, plus legacy Opus 4 and Sonnet 4.

Applications are reviewed monthly on the first Monday, and Anthropic scores them on scientific merit, potential impact, technical feasibility, and a mandatory biosecurity screening. If you’re selected, you must respond within 5 business days, then complete any required compliance affidavits and liability releases within 14 days or you forfeit the award. If you’re not selected, Anthropic typically won’t notify you. You can reapply in a later monthly cycle.

They’re worth $25,000 of Claude API usage, usable across all Claude models in the Anthropic API. Anthropic’s own estimate says that at Claude Sonnet 4.5 rates, $25K is roughly 8.3 billion input tokens or about 1.67 billion output tokens; Haiku 4.5 stretches further to about 25 billion input tokens. In practice, that’s enough budget for serious evaluation work, a real beta, and an early production ramp if you lean on Batch API and caching. The bigger hidden value, honestly, is the priority rate limits if your app needs throughput.

No.

Credits typically expire 12 months from the grant date, and the expiration date can’t be extended.

Yes. If you have Anthropic credits you won’t use before they expire, you can list them on AI Credit Mart and sell them at up to 70% of face value. Companies regularly list surplus credits from startup programs and enterprise agreements.

AI Credit Mart has discounted Anthropic credits available from companies with surplus allocations. Prices are typically 30-70% below retail.

They become unusable, and Anthropic does not extend expiration dates.

The Anthology Fund is a venture investment (Menlo typically invests $100K+ for equity) plus $25,000 in Claude API credits, and it’s highly competitive. The VC Partner Program is credits-only and you apply through your VC’s unique link or referral code, with amounts that can vary (often from about $1,000 up to $25,000). If you don’t want dilution, the partner path is usually the one to pursue.

No. Anthropic uses a prepaid credit model, so API access stops when you hit zero. Auto-reload exists, but it’s opt-in only, meaning you must explicitly enable it.

They’re worth up to $300,000 in promotional credits that automatically offset eligible AWS usage. In practice, that can cover large chunks of Bedrock model usage (including third-party models), EC2 GPU capacity (P5/G5), and Trainium/Inferentia workloads, plus the supporting stack like S3 and databases. The “worth it” part depends on whether you actually run compute-heavy training or inference over the next two years. If your product doesn’t need that level of infrastructure, you’ll leave value unused.

Yes. The AWS account setup step includes confirming a credit card.

These Activate credits expire 2 years from activation, and AWS won’t extend them.

Yes. If you have AWS credits you won’t use before they expire, you can list them on AI Credit Mart and sell them at up to 70% of face value. Companies regularly list surplus credits from startup programs and enterprise agreements.

AI Credit Mart has discounted AWS credits available from companies with surplus allocations. Prices are typically 30-70% below retail.

After expiration, you’re charged standard pay-as-you-go rates for any running services. AWS sends reminder emails around 60 days before, 30 days before, and after expiry, but it won’t pause your infrastructure for you.

Not exactly. Credits do not stack: if you already received $100K in Portfolio credits and later qualify for the Generative AI tier, AWS grants the difference (an additional $200K) to reach the higher allocation.

Using a personal email (Gmail/Yahoo) and looking like an API wrapper are the big ones. AWS also flags vague product/workload descriptions, missing or invalid Org IDs, and prior Activate credits at an equal or greater value. One more: your Builder ID email cannot be tied to another Builder ID, or AWS rejects it as a duplicate. Treat the application like a technical spec, not a branding exercise.

Up to $100,000 in AWS Activate Credits can offset eligible AWS usage across 200+ services (compute, databases, storage, analytics, and AI/ML). On top of that, the package can include up to $10,000 in AWS Business Support credits, 80 self-paced training lab credits, and access to $800K+ in partner offers. The practical value depends on what you run, but it’s enough to support real production infrastructure if you plan usage before the expiration window. One standout: third-party foundation models on Amazon Bedrock (Anthropic, Meta, Mistral, Cohere, Stability, and more) are explicitly covered, which can save a lot if you’re doing serious inference.

Yes, your AWS account must be on a Paid Tier Plan (not free tier).

Portfolio Package credits expire in 1–2 years from issuance, and the exact duration depends on your provider arrangement.

Yes. If you have AWS credits you won’t use before they expire, you can list them on AI Credit Mart and sell them at up to 70% of face value. Companies regularly list surplus credits from startup programs and enterprise agreements.

AI Credit Mart has discounted AWS credits available from companies with surplus allocations. Prices are typically 30-70% below retail.

Unused credits are permanently forfeited, and standard AWS rates kick in immediately for any running services.

Yes. Activate Credits explicitly cover third-party foundation models on Amazon Bedrock, including providers like Anthropic (Claude), Meta (Llama), Mistral, Cohere, AI21 Labs, Stability AI, and Amazon Titan.

Yes, but only in a specific scenario. If you previously received a smaller amount (for example $10,000) and you now qualify for $100,000, AWS allows you to reapply and receive the difference. That means you would be considered for the remaining balance rather than “double dipping” the full amount again. If you previously received credits of equal or greater value, you’re not eligible for this package.

Up to $500,000 total, typically split into about $200K for general AWS services and about $300K for AI silicon and AI usage (Trainium/Inferentia, Bedrock, and reserved H100 capacity). In real usage, that can fund a full AWS stack plus meaningful LLM inference or training, as long as you stay inside eligible services. The exact split may vary by YC batch because AWS and YC negotiate it.

Yes.

These Activate credits last 2 years from activation (when they’re applied to your AWS account). They cannot be extended, and unused credits are forfeited when they expire.

Yes. If you have AWS credits you won’t use before they expire, you can list them on AI Credit Mart and sell them at up to 70% of face value. Companies regularly list surplus credits from startup programs and enterprise agreements.

AI Credit Mart has discounted AWS credits available from companies with surplus allocations. Prices are typically 30-70% below retail.

They’re forfeited, and you can’t get an extension.

Yes. AWS Activate credits can be used on all models available through Amazon Bedrock, including third-party foundation models from providers like Anthropic, Meta, Mistral AI, Cohere, AI21 Labs, and others listed by AWS. Frankly, this is one of the biggest upgrades versus older rules, because before April 2024 AWS credits couldn’t be used for third-party Bedrock models.

No, AWS Marketplace purchases are not covered (except third-party Bedrock models).

ChatGPT Edu isn’t a fixed dollar credit; it’s institution-funded access to ChatGPT features and models. The value comes from replacing an individual subscription and giving you advanced tools (Study Mode, Advanced Data Analysis, agent features) under university privacy controls. Some models also come with clear message limits (like o3 at 100 messages per week), so “worth” depends on how you use those quotas. If you’re doing heavy daily work, the practical value can be comparable to a paid plan, but with added admin controls and data not used for training.

No.

For ChatGPT Edu, access lasts as long as your university’s contract with OpenAI is active, and you remain eligible through enrollment/affiliation.

Yes. If you have OpenAI credits you won’t use before they expire, you can list them on AI Credit Mart and sell them at up to 70% of face value. Companies regularly list surplus credits from startup programs and enterprise agreements.

AI Credit Mart has discounted OpenAI credits available from companies with surplus allocations. Prices are typically 30-70% below retail.

With ChatGPT Edu specifically, you don’t “expire credits”; you lose access when your institution’s contract ends or when you’re no longer enrolled.

Because your workspace admin controls which models are enabled, and some options are disabled by default. GPT-5.2 is listed as available in early access but requires admin enablement, and GPT-5.1 is also described as disabled by default. If you need a specific model for a class or lab, your best move is to ask IT or the workspace owner what the institution’s policy is and whether they can enable it for your group.

Limits vary by model, and they apply per user. The program notes examples like o3 at 100 messages per week, o4-mini at 300 messages per day, and o4-mini-high at 100 messages per day. When you hit a limit, that model is temporarily grayed out until the reset time.

They’re worth up to $250,000 of Cloudflare pay-as-you-go usage over 1 year, but two services have hard caps: R2 + Cache Reserve tops out at $10,000, and Workers AI tops out at $50,000. Everything else draws from the remaining balance (up to about $190,000). In practice, the value is biggest if you’re running real production workloads on Workers plus security/performance features for up to three domains. If you were planning to pay for WAF, DDoS protection, and Zero Trust anyway, the “included Enterprise-level domains” portion can be a quiet bargain.

Yes. The signup flow explicitly requires adding a valid credit card to your Cloudflare account.

These startup credits expire after 1 year (or when you use them up).

Yes. If you have Cloudflare credits you won’t use before they expire, you can list them on AI Credit Mart and sell them at up to 70% of face value. Companies regularly list surplus credits from startup programs and enterprise agreements.

AI Credit Mart has discounted Cloudflare credits available from companies with surplus allocations. Prices are typically 30-70% below retail.

Your account switches to standard pay-as-you-go billing, and Enterprise domain features revert to your current plan level. There’s no automatic renewal, and you can reapply only if you still meet eligibility (with no guarantee of approval).

R2 + Cache Reserve is capped at $10,000, and Workers AI is capped at $50,000, regardless of the $250,000 headline amount.

No. The program explicitly excludes Cloudflare Registrar, so domain purchases must be paid separately.

They’re worth $100,000 of Cloudflare usage across the Developer Platform plus enterprise features on up to three domains, with specific caps of $50K for Workers AI and $10K combined for R2 + Cache Reserve.

Yes. Cloudflare requires a credit card on file, but it won’t be charged until credits run out (or the year ends and you keep using paid services).

The Seed credits last 1 year or until fully consumed, whichever comes first.

Yes. If you have Cloudflare credits you won’t use before they expire, you can list them on AI Credit Mart and sell them at up to 70% of face value. Companies regularly list surplus credits from startup programs and enterprise agreements.

AI Credit Mart has discounted Cloudflare credits available from companies with surplus allocations. Prices are typically 30-70% below retail.

Your Startup Plan ends, usage-based services move to pay-as-you-go billing on your card, and enterprise domain features revert to your account’s plan level. There’s no automatic cancellation warning, so you need billing alerts. If you still need help, Cloudflare says you can request additional credits case-by-case via startups@cloudflare.com, but approval isn’t guaranteed.

Workers AI is capped at $50,000, and R2 + Cache Reserve share a combined $10,000 cap.

Yes, but you’ll likely need to email startups@cloudflare.com with your perk manager’s details so Cloudflare can try to establish a partnership.

Up to $50,000 in credits, applied to Databricks platform usage measured in DBUs. In practice, that can fund interactive development (notebooks and all-purpose clusters), scheduled production pipelines on Jobs Compute, SQL Warehouses for analytics, and Mosaic AI Model Serving for deploying models as APIs. It’s not “free cloud,” though. You still pay AWS/Azure/GCP for the underlying VMs, storage, and networking, so the true runway extension depends on how efficiently you design clusters and how heavy your data movement is.

No for the application itself, but you will need an AWS, Azure, or GCP account to use Databricks.

Databricks does not publicly disclose the validity period for the startup program credits, so you should confirm expiration during the application.

Yes. If you have Databricks credits you won’t use before they expire, you can list them on AI Credit Mart and sell them at up to 70% of face value. Companies regularly list surplus credits from startup programs and enterprise agreements.

AI Credit Mart has discounted Databricks credits available from companies with surplus allocations. Prices are typically 30-70% below retail.

You’ll need to set up billing to keep using Databricks after credits are exhausted, and Databricks notes there’s no automatic conversion to paid billing without your consent.

No. The credits cover Databricks platform charges only; your cloud provider bills compute, storage, and networking separately.

Yes, it’s described as available globally, but you still need access to Databricks on AWS, Azure, or GCP. So if your org can’t open a cloud account (or can’t meet requirements like GCP’s org and billing setup), you’ll hit friction even if the program itself has no region restriction.