Our External Researcher Access Program is specifically designed to support and lower the barrier to entry for researchers working on AI safety and alignment topics that we consider high priority by providing free API credits.
If you are an alignment researcher and are interested in receiving free API credits for AI safety and alignment research purposes, you might qualify for our External Researcher Access Program.
Please complete the following application form with details about your team and research topic:
About this Program
Please note that you can access the Claude model family immediately via API by simply creating an Anthropic Console account – you do not need to contact us first.
This program provides free API credits for our standard model suite to researchers working on AI safety and alignment topics we consider high priority.
Note that these credits are for API use and do not apply to Claude.ai.
This program does not provide access to non-standard or non-public models.
Applicants through this program do not receive exemption from our Usage Policy. Our Trust & Safety team will follow our standard enforcement procedures and take action whenever an organization’s prompt activity hits our violation rate thresholds.
About our Process
We evaluate submissions on the 1st Monday of each month. This schedule helps us manage the program sustainably. Please note that if this timeline poses a significant obstacle for the applicant, it does not block them from simply purchasing API credits in the interim.
If successful, we will apply $1000 in API credits to the account. For rare special cases, we may approve a higher quantity of credit.
Please note that given the substantial number of applications we receive (sometimes thousands in a single week), we regret that we cannot provide individual responses to unapproved submissions. In addition, we reserve the right to reject submissions for any reason and will not be able to provide explanations for each individual applicant. However, we appreciate the time and effort put into each submission and will carefully review all entries. Thank you for your understanding.
If you are having an issue that requires a response urgently (e.g., you’ve suddenly run out of your credits before a major conference), please email researcheraccess@anthropic.com. However, please note that we will not be able to respond if your question is already addressed by the following FAQ.
Frequently Asked Questions
I would like access to non-standard or nonpublic models for my research. Is this possible?
We understand the potential value this could bring to your work, however, to balance supporting valuable research with maintaining the safety and integrity of our systems, we currently have the following policy in place:
To maintain the sustainability of this free, high-volume program, we are unable to extend access to any nonpublic or experimental models through this External Researcher Access program. This non-standard or nonpublic model access is reserved for a very limited number of pre-deployment testing partnerships, which are not part of this program.
We do not currently provide finetuning access via an API at this time. Finetuning access to Claude models is in preview mode via AWS Bedrock and not covered via this program.
We continually evaluate our offerings and may consider expanding access to additional models in the future.
I run a research program with many researchers and collaborators. What's the best way for us to apply for free API credits via this program?
For research programs with multiple researchers and collaborators, we recommend setting up a single account to represent your entire program, and submit an application through our application form: Application Form - Anthropic's External Researcher Access Program
If approved, we'll allocate credits to a single organization ID, which you can then flexibly distribute among your researchers. This enables you to easily add or remove collaborators as needed, monitor usage across your team, and ensure compliance with our policies.
If your program's needs change significantly over time, please submit an updated application to discuss potential adjustments to your allocation.
I am performing red-teaming or jailbreaking research. Is there a way to make sure our account does not get flagged for potentially breaking your Usage Policy while our researchers are purposefully red-teaming?
We appreciate your interest in conducting red-teaming or jailbreaking research, as it plays a crucial role in improving AI safety. However, to maintain the sustainability of our External Researcher Access Program, participants are not exempt from our Usage Policy. Our Trust & Safety team will follow our standard enforcement procedures and take action whenever an organization’s prompt activity hits our violation rate thresholds.
However, to balance supporting valuable research with maintaining the safety and integrity of our systems, we have created two alternative opportunities:
Close collaborators of Anthropic employees: If you are already collaborating closely with an Anthropic employee on your research and require an exemption to our Usage Policy, please inform your Anthropic collaborator about your needs. They can initiate our internal approval process, and, if approved via this process, your account will be appropriately labeled.
Model Safety Bug Bounty Program: For researchers focusing specifically on jailbreaking, we encourage you to apply to our Model Safety Bug Bounty Program. This program is designed to support and reward research in this area.
Does your Researcher Access Program provide free API credits for other areas of research?
Our Researcher Access Program is specifically designed to support and lower the barrier to entry for researchers working on AI safety and alignment topics that we consider high priority. To maintain the sustainability of this free, high-volume program, we are currently focusing our resources on these specific areas.
While we're unable to extend free API credits to other fields of research at this time, we want to emphasize that this doesn't prevent you from accessing our API for your research purposes. We welcome researchers from all disciplines to purchase API credits and conduct their studies using our platform. We value the diverse range of research being conducted with our API and are always excited to hear about the innovative ways it's being used across various fields.
I am a researcher based in a country where Claude is currently unavailable. Can I access your API through this program?
We greatly appreciate your interest in using Claude for your research. However, we are unable to provide API access in countries where Claude is not yet officially available.
We understand this may be disappointing. In the meantime, we encourage you to stay informed about Claude's availability updates through our news updates.
I have contacted you about free API credits through your Researcher Access Program and have not heard back about my application. Who should I contact?
Thank you for your interest in our Researcher Access Program and for taking the time to submit an application. We appreciate your patience. Here's what you need to know about our application process:
High volume of applications: We receive thousands of applications each month. To maintain the sustainability of this free, high-volume program, we are unable to provide individual responses to all applicants.
Monthly evaluation schedule: We evaluate submissions on the 1st Monday of each month. This schedule helps us manage the program sustainably and ensure fair consideration for all applicants.
Other options: If our monthly evaluation schedule poses a significant obstacle for your research timeline, please note that you're welcome to purchase API credits in the interim to begin your work immediately.
Approval notifications:
If your application is approved, you will be notified promptly after the monthly evaluation.
If you haven't heard back from us, it means that your application has not been approved.
Reapplication: Please note that this program was substantially revised in July 2024. If you contacted us before that time, or if significant time has passed since your initial application, you're welcome to submit a new application through the new application form with any updated information about your research: Application Form - Anthropic's External Researcher Access Program
Thank you for your understanding. We're committed to supporting the AI research community to the best of our ability while maintaining a sustainable process.