Free Claude API access: what is actually free.

Quick answer: four routes, four catches.

There is no self-hosting route at all: Anthropic has never published Claude weights, so unlike the open-weight Chinese families the free-Claude question is entirely a hosted-access question. The four routes above are the complete list, and only two of them end with a key that calls claude ids.

• Claude.ai's free chat tier: genuinely free daily usage of Claude in a browser or app. The catch: it is a consumer chat product, no API key comes with it, and paying for a Claude.ai subscription does not add API credits either.
• Anthropic console trial credit: 2026 guides describe a small one-time credit (about 5 USD) for new console accounts after phone verification, and other write-ups call it unreliable or absent. The catch: it is a trial at best, not a tier, and the console is otherwise prepaid from the first request.
• Eligibility-gated credit programs: Anthropic runs startup and open-source credit programs that guides describe as reaching substantial amounts. The catch: each has an application, criteria, and a review; none is a sign-up-and-go free API.
• APIsRouter's free starting credit: issued without a card, spendable on claude-sonnet-4-6, claude-opus-4-7, claude-haiku-4-5-20251001, and claude-sonnet-5 at the same catalog rates as paid traffic. The catch is the honest one: a starting credit, not an unlimited tier; pay-as-you-go after.

What Anthropic offers, and what only looks like an offer.

The developer platform is prepaid pay-as-you-go: create an account, buy credits, generate a key, and every request bills per token at the official list rates. That has been the shape for years, and nothing checked in July 2026 changes it: there is no free model list, no free daily quota, and no card-free path to sustained usage on Anthropic's own console. Around that paid core sit the things that get remembered as a free tier. New console accounts have at times received a small trial credit, around 5 USD, gated behind phone verification; 2026 guides disagree on whether it is currently granted, so treat it as a pleasant surprise rather than a plan. The startup, research, and open-source credit programs are real but application-based. And Claude.ai subscriptions, from the free tier up through the paid plans, are a separate product line entirely: none of them includes API credits, and API spend does not count against any chat plan. The distinction that resolves most confusion: Anthropic sells chat access and API access as two unrelated meters. A free route only answers the search in this page's title if it ends with an sk-style key answering at an HTTPS endpoint.

What the free starting credit here actually is.

The route on this site is deliberately simple: a free starting credit, no card required, spendable on any catalog id, including the full Claude family. There is no degraded free endpoint and no separate free-model list; the credit meters the same models at the same catalog rates as paid traffic. When it runs out, access continues as pay-as-you-go top-ups with no subscription, and an idle month costs nothing. To be precise about what this is not: it is not unlimited free Claude. Nobody serves a frontier model without metering it somewhere, and any page promising otherwise is describing a rate limit in optimistic language. What the credit buys is a real evaluation: current models, real rate behavior, and exact per-request token counts in the usage log, before any money moves. The practical pattern: start on claude-sonnet-4-6, watch what your prompts actually cost in the console, and let those numbers decide whether Haiku volume pricing or Opus depth is the right steady state. The same key also runs DeepSeek, GLM, and GPT ids, so the "is Claude worth its rate for this workload" question becomes an afternoon of identical requests rather than a forum debate.

curl https://api.apisrouter.com/v1/chat/completions \
  -H "Authorization: Bearer $APISROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4-6",
    "messages": [{"role": "user", "content": "Draft three unit tests for a URL parser, as a table."}]
  }'

Claude.ai's free tier, and why it keeps fooling people.

Claude.ai has a real free tier: daily-limited chat with current Claude models, no payment required. It is the right answer for trying Claude conversationally and the wrong answer for every search that ends in "API", because no automation can sit on top of it, and scraping the chat UI violates the terms it runs under. The confusion is structural: one brand covers a free consumer product and a prepaid developer platform, and coverage of one gets read as coverage of the other. The test is a single question: does the route hand you a key that answers at an endpoint? The chatbot does not, at any plan level. The routes that do, Anthropic's prepaid console and gateway credits like the one here, are the API answers, and each meters usage because serving frontier models costs someone money per token. One adjacent case worth naming: Claude Code. The agent bills either against a Claude.ai subscription plan or against API credits, so a subscription can fund agent sessions, but that is still not an API key you can point your own code at.

Pay-as-you-go · transparent per-model pricing

Selected models are priced below official list prices. Exact input, output, cache, and per-request prices are shown for each model.

When free stops being the right answer.

Free Claude routes fail in characteristic ways: the chat tier caps the day's usage mid-conversation, a trial credit is spent in one honest evaluation, an application-based program takes weeks you do not have. The signal to move is the first time a limit costs more time than the metered price would have cost money. The Claude family makes that transition gradual rather than dramatic. claude-haiku-4-5-20251001 carries high-volume workloads at the family's lowest rates, claude-sonnet-4-6 is the general workhorse, and claude-opus-4-7 is priced for the requests that genuinely need it. The pricing table above puts exact per-million rates next to each id; the billing guide linked below works the math for real request shapes. Because the starting credit and paid traffic share one endpoint and one key, outgrowing free involves no migration: the same base URL, the same model strings, and a balance that tops up instead of a cap that resets.

Which free route fits which situation.

• Trying Claude conversationally: Claude.ai's free tier. Free, daily-limited, no API.
• A funded startup or an open-source maintainer: check Anthropic's credit programs; the amounts guides report are worth an application if you qualify.
• Evaluating claude ids for real work, today, without a card: the starting credit here, because it runs the exact ids you would ship with, next to every model you might compare them against.
• A team already on Anthropic's console: check whether the new-account trial credit landed, and treat it as evaluation budget rather than a plan.
• Production traffic of any size: metered pay-as-you-go, on Anthropic's console or a gateway. No free mechanism anywhere is built to carry production, and Claude has no self-host escape hatch.

FAQ