Listen first: the two-minute version
M:SS · The full post is below.

What changed on October 6

For the past year, Anthropic's Claude for Startups program had one rule that shut out most small businesses: you needed outside money. The archived version of the program page, from a week before the expansion, says you had to be "funded by institutional investors" and "founded within the last four years." That's a program for venture-backed startups, not for founders.

On October 6 — the first day of SF Tech Week — Anthropic expanded the program, and TechCrunch reported the new shape: a free year of Claude Team, $1,000 in API credits, and up to $45,000 in partner offers. But the change that matters for a bootstrapped business is eligibility. The program is now, in Anthropic's own words, "open to bootstrapped, pre-seed, and venture-backed startups."

Before (page as of Sept 29, 2026)After (Oct 6, 2026)
Who qualifies VC-backed only: "funded by institutional investors," founded within 4 years Founded in the last 5 years or funded in the last 2 — bootstrapped welcome
API credits Credits offered, amount not published $1,000, first-party Claude API
Claude Team Not included Free year, up to 5 Premium seats
Partner offers Not published Up to $45K through the Claude Startup Stack
Access to Anthropic Priority rate limits 45-minute office hours every other week with the Applied AI team, plus events

The free year is the big jump. Claude Team is the plan that puts Claude Code, Claude in Chrome, and the Microsoft 365 integration on every seat, and Premium seats list at $100–125 a seat a month depending on billing. Five Premium seats for a year is a $6,000–7,500 line item that qualifying businesses now get written off.

Who qualifies now — and what you need

Two caveats. First, "we review every application" — per the official program terms, benefits are granted at Anthropic's sole discretion, and applicants must comply with Anthropic's supportability policies. This is a real program, not a coupon. Second, if you're already a member of the old program, you don't reapply — you're now eligible to redeem the expanded benefits directly.

What $1,000 in API credits actually buys

Credits are the part with real math, so here it is at current list price (per million tokens, input / output): Haiku 4.5 at $1/$5, Sonnet 5.5 at $2/$10, Opus 5.5 at $4/$20, Fable 5.1 at $10/$50.

A typical agent task — about 1,000 input tokens, 500 output tokens: the equivalent of one thorough lead response, one draft, one research pass — costs under half a cent on the fast model and a few cents on the flagship. That's what the same $1,000 stretches to:

ModelPrice (per M tokens, in/out)One task (1k in / 500 out)Tasks $1,000 buys
Haiku 4.5$1 / $5~0.35¢~285,000
Sonnet 5.5$2 / $10~0.7¢~143,000
Opus 5.5$4 / $20~1.4¢~71,000
Fable 5.1$10 / $50~3.5¢~28,500

Even on the most expensive model, that's tens of thousands of agent tasks — months of production volume for a small business running a lead-response agent or an automated content workflow. And if a chunk of your work is non-urgent, Anthropic's Batch API runs the same models at 50% off, which effectively doubles the budget for background work.

The fine print

How to apply — about ten minutes

Who this is really for

Strip away the startup framing and here's what this is: a business founded in the last five years getting $6,000–7,500 in seat value, $1,000 in compute, up to $45,000 in tool discounts, and biweekly access to the people who build Claude — for free, as long as it qualifies.

If you're running a business in 2026 and you're planning to put AI on real work — answering leads, drafting content, running the back office — the year-one cost of a serious AI stack just dropped to near zero. The AI cost breakdown does the math on what AI replaces on a typical local P&L; this is the program that makes the remaining line items nearly free.