Which AI Is Best for Grant Writing in 2026?

Grant writing takes time that small nonprofits, research labs and startups rarely have, and AI tools now promise to find funders, draft narratives and build budgets in a fraction of the time. Grant databases such as Candid now use AI to recommend funders to nonprofits. Funders are paying attention too. The National Institutes of Health announced in 2025 that it will not consider applications substantially developed by AI to be original work. To find out which AI is best for grant writing, our research team ran a head-to-head study from September 29 to October 16, 2026.

We gave the same 480 grant writing requests to nine AI platforms: seven general AI chatbots and two grant tools, Grantable and Instrumentl, producing 4,320 answers. The requests covered eight tasks, with 60 requests each, split evenly across nonprofit foundation grants, federal research grants and small business grants, and were built from 30 real organizations’ programs and budgets. A panel of six experienced grant reviewers and development directors, including former NIH and NSF panel reviewers and foundation program officers, scored every answer from 0 to 100 using each funder’s own scoring criteria, without knowing which platform wrote it. We also checked every draft for compliance with the funder’s rules, verified every statistic and citation, and confirmed whether each of the 3,912 grant opportunities the platforms recommended was real and open.

Best AI for Grant Writing Overall

Each platform was tested on its best available model or plan. A platform’s overall score is the average of its scores across the eight tasks in our study, with each task weighted equally. In the table below, we rank the nine platforms from highest to lowest.

Grant Writing Scores by AI Platform (September to October 2026)

Rank	AI Platform	Model or Plan Tested	Overall Score	Strongest Task
1	Claude	Claude Opus 5.5	89.6	Project Narratives
2	ChatGPT	GPT-6 Sol	89.5	Project Narratives
3	Grantable	Grantable AI assistant (paid plan)	87.7	Following Funder Guidelines
4	Google Gemini	Gemini 4 Argon	87.5	Budgets and Justifications
5	Instrumentl	Instrumentl AI tools (paid plan)	84.9	Finding Matching Grants
6	Microsoft Copilot	GPT-6 Astra	84.4	Budgets and Justifications
7	Grok	Grok 4.6	81.5	Needs Statements
8	DeepSeek	DeepSeek V4-Pro	80.0	Budgets and Justifications
9	Meta AI	Muse Spark	78.4	Needs Statements

Our researchers took three main points from this data:

• We found that Claude and ChatGPT are close at the top, separated by 0.1 points. Claude finished first or second in six of the eight tasks, and reviewers said its narratives read the most like proposals written by an experienced grant writer.
• Grantable was the highest-ranked grant tool, in third place at 87.7. It works from the documents an organization uploads, such as past proposals and annual reports, which helped it follow each funder’s instructions closely.
• Instrumentl ranked fifth overall but led both tasks about finding funding. Its writing scores were lower, from 79.6 to 82.4 on narratives, needs statements, budgets and logic models.

Best AI for Each Grant Writing Task

We split the requests evenly across eight tasks, with 60 requests per task. The first two tasks cover finding funding, and the other six cover writing and preparing the application. The first table below shows the winner and runner-up for each task. The second table and the heat map that follows show every platform’s score for every task, with the top score in each column in bold.

Best AI Platform for Each Grant Writing Task (September to October 2026)

Grant Writing Task	Best Platform	Score	Runner-Up	Score
Finding Matching Grants	Instrumentl	92.6	Google Gemini	87.9
Checking Eligibility	Instrumentl	91.3	ChatGPT	89.1
Needs Statements	Claude	91.8	ChatGPT	90.6
Project Narratives	Claude	92.7	ChatGPT	90.9
Budgets and Justifications	ChatGPT	90.3	Claude	89.4
Logic Models and Evaluation Plans	Claude	91.2	ChatGPT	89.7
Following Funder Guidelines	Grantable	91.4	Claude	90.8
Letters of Inquiry	Claude	91.6	ChatGPT	90.8

Grant Writing Scores by Task and AI Platform (September to October 2026)

AI Platform	Finding Grants	Eligibility	Needs Statements	Narratives	Budgets	Logic Models	Guidelines	Letters of Inquiry
Claude	80.4	88.6	91.8	92.7	89.4	91.2	90.8	91.6
ChatGPT	85.4	89.1	90.6	90.9	90.3	89.7	88.9	90.8
Grantable	83.6	86.9	88.4	88.9	86.2	87.8	91.4	88.7
Google Gemini	87.9	88.2	87.6	87.4	88.7	86.9	86.8	86.3
Instrumentl	92.6	91.3	82.4	81.7	80.9	79.6	85.7	84.8
Microsoft Copilot	82.1	84.6	84.2	84.8	87.2	83.9	84.4	83.7
Grok	79.8	81.2	83.6	82.9	81.4	80.7	79.6	83.1
DeepSeek	71.4	79.6	81.8	82.4	83.2	81.6	78.9	80.8
Meta AI	74.2	77.8	82.1	80.6	76.9	77.4	76.2	81.9

Here is what our team took from the task-level results:

• We found that Claude won four tasks, Instrumentl won two, and ChatGPT and Grantable won one each. ChatGPT led budgets and justifications (90.3), with the clearest line-by-line explanations of each cost.
• Finding matching grants produced the widest gap of any task, 21.2 points between Instrumentl (92.6) and DeepSeek (71.4). Instrumentl and Google Gemini (87.9) searched current funder data, while chatbots without that data often suggested programs from past years.
• Grantable led following funder guidelines (91.4). Claude scored 80.4 on finding matching grants, where it ranked sixth, so it works best once an organization knows which grant it is applying for.

Accuracy: Real Grants, Funder Rules and Invented Facts

A strong draft can still be rejected without review if it breaks a funder’s rules, and an invented statistic can damage an organization’s credibility with a funder. We confirmed whether each of the 3,912 recommended grant opportunities was real and accepting applications, checked every draft against the funder’s page limits, word limits and required sections, and verified every statistic and citation. The table below is sorted from the fewest to the most invented facts.

Grant Writing Accuracy by AI Platform (September to October 2026)

AI Platform	Recommended Grants That Were Real and Open	Proposals With a Compliance Error	Proposals With an Invented Statistic or Citation
Claude	78.6%	9.4%	3.1%
Grantable	93.4%	6.8%	3.8%
Instrumentl	97.2%	12.6%	4.6%
ChatGPT	86.9%	11.2%	5.2%
Google Gemini	90.8%	13.1%	5.9%
Microsoft Copilot	82.3%	15.8%	7.4%
Grok	69.4%	21.4%	12.6%
Meta AI	61.8%	27.9%	14.8%
DeepSeek	54.2%	24.7%	16.3%

Our researchers drew three conclusions from the accuracy data:

• We found that Claude invented a statistic or citation in 3.1% of proposals, followed by Grantable at 3.8%. DeepSeek did so in 16.3% of proposals, about one in every six.
• Grantable had the fewest compliance errors (6.8%), and Meta AI had the most (27.9%). The most common error was going over a page or word limit, followed by leaving out a section the funder required.
• Instrumentl recommended the most real, open grants (97.2%). Of the grants DeepSeek recommended, 45.8% were closed, out of date or did not exist.

Best AI for Each Type of Grant

Nonprofit foundation grants, federal research grants and small business grants follow different formats and are judged on different criteria. Each type made up 160 of our 480 requests. The table below shows each platform’s average score for each type of grant, with the top score in each column in bold.

Grant Writing Scores by Type of Grant and AI Platform (September to October 2026)

AI Platform	Nonprofit Foundation Grants	Federal Research Grants (NIH and NSF)	Small Business Grants (SBIR and STTR)
Claude	89.2	90.6	88.9
ChatGPT	89.4	89.1	90.1
Grantable	90.2	85.4	87.5
Google Gemini	87.1	88.3	87.0
Instrumentl	86.8	82.6	85.2
Microsoft Copilot	84.1	83.9	85.2
Grok	81.9	80.6	82.1
DeepSeek	79.2	81.4	79.3
Meta AI	79.8	76.4	78.9

Here is what our team took from the results by type of grant:

• We found that a different platform led each type of grant. Grantable led nonprofit foundation grants (90.2), Claude led federal research grants (90.6) and ChatGPT led small business grants (90.1).
• Both grant tools scored lowest on federal research grants: Grantable dropped to 85.4 and Instrumentl to 82.6. Reviewers said their drafts were weaker on the research strategy and scientific background these applications require.
• Federal funders have set rules on AI. NIH limits each principal investigator to six applications per calendar year and will not treat applications substantially developed by AI as original work, and NSF encourages applicants to say which parts of a proposal were made with AI.

Best AI Models and Plans for Grant Writing, Including Free Options

Many nonprofits work with small budgets, so free options matter. We tested 13 models and plans across the nine platforms, including each platform’s free option where one exists. The table below ranks every one we tested.

Grant Writing Scores by Model and Plan (September to October 2026)

Rank	Model or Plan	Platform	Available On	Score
1	Claude Opus 5.5	Claude	Paid	89.6
2	GPT-6 Sol	ChatGPT	Paid	89.5
3	Claude Sonnet 5.5	Claude	Free and paid	88.7
4	Grantable (paid plan)	Grantable	Paid	87.7
5	Gemini 4 Argon	Google Gemini	Paid	87.5
6	GPT-6 Luna	ChatGPT	Free and paid	87.1
7	Gemini 3.8 Flash	Google Gemini	Free and paid	86.4
8	Instrumentl (paid plan)	Instrumentl	Paid	84.9
9	Grantable (free plan)	Grantable	Free	84.6
10	GPT-6 Astra	Microsoft Copilot	Free and paid	84.4
11	Grok 4.6	Grok	Free and paid	81.5
12	DeepSeek V4-Pro	DeepSeek	Free	80.0
13	Muse Spark	Meta AI	Free	78.4

Our team took three main points from the model-level results:

• We found that Claude Sonnet 5.5 is the best free AI for grant writing, scoring 88.7, behind only Claude Opus 5.5 and GPT-6 Sol.
• Grantable’s free plan scored 84.6, 3.1 points below its paid plan and higher than the free options from Microsoft Copilot, Grok, DeepSeek and Meta AI.
• Instrumentl has no free plan, but its paid plan led both tasks about finding grants, so it can work well alongside a chatbot that handles the writing.

Requesting a Copy of This Report

If you’d like to request a PDF copy of this report or learn more about our agency, you can reach out here.