People now bring AI chatbots their questions about relationships, money, health, work and family. In a J.D. Power survey reported by NPR in August 2026, 40% of people said they had used AI to help manage their finances in the prior three months. To find out which AI is best for advice, our research team ran a head-to-head study from September 29 to October 16, 2026.
We asked seven AI platforms the same 520 advice questions, producing 3,640 answers. The questions covered eight areas of life, with 65 questions each. A panel of 16 licensed professionals, two for each area, including therapists, physicians, attorneys, certified financial planners, career coaches and pediatricians, scored each answer from 0 to 100 without knowing which platform wrote it. We also ran 120 prompts describing a personal crisis and 80 prompts in which the user was clearly in the wrong, to measure how each platform handled safety and how often it simply agreed with the user.
Summary of Findings
- Claude is the best AI for advice overall, scoring 89.8 out of 100 with Claude Opus 5.5, followed by ChatGPT at 89.2 and Google Gemini at 87.3.
- Claude won four of the eight areas of life we tested, including emotional support (92.7) and relationships and family (91.8). ChatGPT led career and everyday decisions, and Gemini led health and personal finance.
- Claude recommended a doctor, lawyer, therapist or other professional in 94.4% of cases where one was needed and pointed users to crisis help in 98.3% of crisis prompts, the highest rates in the study.
- Meta AI sided with users who were clearly in the wrong 41.2% of the time, more than twice Claude’s rate of 18.8%.
Best AI for Advice Overall
Each platform was tested on its best available model. A platform’s overall score is the average of its scores across the eight areas of life in our study, with each area weighted equally. In the table below, we rank the seven platforms from highest to lowest.
Advice Scores by AI Platform (September to October 2026)
| Rank | AI Platform | Model Tested | Overall Score | Strongest Area |
| 1 | Claude | Claude Opus 5.5 | 89.8 | Emotional Support |
| 2 | ChatGPT | GPT-6 Sol | 89.2 | Career and Workplace |
| 3 | Google Gemini | Gemini 4 Argon | 87.3 | Health and Wellness |
| 4 | Microsoft Copilot | GPT-6 Astra | 85.1 | Career and Workplace |
| 5 | Grok | Grok 4.6 | 82.1 | Everyday Decisions |
| 6 | Meta AI | Muse Spark | 81.4 | Emotional Support |
| 7 | DeepSeek | DeepSeek V4-Pro | 80.3 | Everyday Decisions |
Our researchers took three main points from this data:
- We found that Claude led ChatGPT by 0.6 points and was the only platform to score 90 or higher in three areas. The panel noted that it asked follow-up questions about the user’s situation more often than any other platform.
- Google Gemini ranked third overall but led health and wellness (90.3) and personal finance (89.2), the two areas where accurate facts mattered most to the panel.
- Meta AI ranked sixth overall but placed third for emotional support (85.3), where judges rewarded its warm, conversational tone.
Best AI for Each Area of Life
We split the questions evenly across eight areas of life, with 65 questions per area. The first table below shows the winner and runner-up in each area. The second table and the heat map that follows show every platform’s score in every area, with the top score in each column in bold.
Best AI Platform for Each Area of Life (September to October 2026)
| Area of Life | Best Platform | Score | Runner-Up | Score |
| Relationships and Family | Claude | 91.8 | ChatGPT | 89.6 |
| Career and Workplace | ChatGPT | 90.4 | Claude | 89.7 |
| Personal Finance | Google Gemini | 89.2 | ChatGPT | 88.9 |
| Health and Wellness | Google Gemini | 90.3 | ChatGPT | 88.6 |
| Legal Questions | Claude | 89.6 | ChatGPT | 88.2 |
| Parenting | Claude | 90.4 | ChatGPT | 89.7 |
| Emotional Support | Claude | 92.7 | ChatGPT | 88.4 |
| Everyday Decisions | ChatGPT | 89.8 | Claude | 88.9 |
Advice Scores by Area of Life and AI Platform (September to October 2026)
| AI Platform | Relationships | Career | Finance | Health | Legal | Parenting | Emotional Support | Everyday |
| Claude | 91.8 | 89.7 | 87.3 | 88.1 | 89.6 | 90.4 | 92.7 | 88.9 |
| ChatGPT | 89.6 | 90.4 | 88.9 | 88.6 | 88.2 | 89.7 | 88.4 | 89.8 |
| Google Gemini | 85.9 | 87.6 | 89.2 | 90.3 | 87.4 | 86.8 | 84.2 | 87.1 |
| Microsoft Copilot | 83.2 | 88.1 | 86.4 | 85.7 | 84.9 | 84.3 | 82.6 | 85.4 |
| Grok | 82.7 | 83.5 | 82.1 | 80.4 | 81.3 | 79.6 | 80.8 | 86.2 |
| Meta AI | 84.1 | 80.2 | 78.6 | 77.9 | 76.8 | 83.5 | 85.3 | 84.7 |
| DeepSeek | 79.8 | 81.9 | 81.4 | 79.6 | 80.7 | 78.9 | 77.4 | 82.6 |
Here is what our team took from the results by area of life:
- We found that Claude won four areas, while ChatGPT and Google Gemini won two each. No other platform finished first in any area.
- Emotional support produced the widest gap of any area, 15.3 points between Claude (92.7) and DeepSeek (77.4). Everyday decisions produced the narrowest gap at 7.2 points.
- Grok’s best area was everyday decisions (86.2), where it placed fourth. Microsoft Copilot placed third for career and workplace advice (88.1).
Safety: Professional Referrals, Crisis Responses and Unsafe Advice
Some questions call for more than a chatbot can offer. Our panel flagged 160 of the 520 questions as ones where a good answer should recommend a doctor, lawyer, therapist, financial advisor or other professional, and we counted how often each platform did. We also measured how often each platform pointed users to a crisis line or emergency help across the 120 crisis prompts, and the share of all answers the panel rated unsafe or factually wrong.
Safety Results by AI Platform (September to October 2026)
| AI Platform | Recommended a Professional When Needed | Pointed to Crisis Help | Unsafe or Wrong Advice |
| Claude | 94.4% | 98.3% | 2.1% |
| ChatGPT | 92.5% | 97.5% | 2.9% |
| Google Gemini | 93.1% | 96.7% | 2.5% |
| Microsoft Copilot | 90.6% | 95.0% | 3.7% |
| Grok | 78.1% | 85.8% | 7.9% |
| Meta AI | 81.9% | 90.8% | 6.9% |
| DeepSeek | 76.9% | 83.3% | 8.5% |
Our researchers drew three conclusions from the safety data:
- We found that Claude recommended a professional in 94.4% of the cases where one was needed, followed by Google Gemini at 93.1% and ChatGPT at 92.5%.
- DeepSeek and Grok gave unsafe or wrong advice in 8.5% and 7.9% of answers, roughly four times Claude’s rate of 2.1%. Most of their errors came in health and legal questions.
- Claude pointed users to a crisis line or emergency help in 98.3% of crisis prompts. DeepSeek did so in 83.3%, missing about one in every six.
How Often AI Just Agrees With You
Good advice sometimes means telling people what they do not want to hear. In 80 of our prompts, the user described a conflict in which, by the panel’s judgment, the user was clearly in the wrong, such as reading a partner’s messages or skipping a friend’s wedding without notice. We counted how often each platform sided with the user anyway. The table below is sorted from least to most likely to side with the user.
How Often AI Sided With Users Who Were in the Wrong (September to October 2026)
| AI Platform | Prompts Where It Sided With the User | Rate |
| Claude | 15 of 80 | 18.8% |
| Google Gemini | 20 of 80 | 25.0% |
| ChatGPT | 22 of 80 | 27.5% |
| Microsoft Copilot | 24 of 80 | 30.0% |
| Grok | 27 of 80 | 33.8% |
| DeepSeek | 29 of 80 | 36.2% |
| Meta AI | 33 of 80 | 41.2% |
Here is what our team took from these results:
- We found that Meta AI sided with the user in 41.2% of these prompts, the highest rate in the study, while Claude had the lowest rate at 18.8%.
- Every platform sided with the user in at least one of every six of these prompts. This matches a March 2026 Stanford study in Science, which found that chatbots affirmed users’ actions 49% more often than humans did.
- Google Gemini (25.0%) and ChatGPT (27.5%) were the next least likely to agree with a user who was in the wrong, and both most often gently pointed out the other person’s view.
Best AI Models for Advice, Including Free Options
Most platforms give free users a different model than paying subscribers. We tested 11 models across the seven platforms, including each platform’s free option. The table below ranks every model we tested.
Advice Scores by Model (September to October 2026)
| Rank | Model | Platform | Available On | Score |
| 1 | Claude Opus 5.5 | Claude | Paid | 89.8 |
| 2 | GPT-6 Sol | ChatGPT | Paid | 89.2 |
| 3 | Claude Sonnet 5.5 | Claude | Free and paid | 88.6 |
| 4 | Gemini 4 Argon | Google Gemini | Paid | 87.3 |
| 5 | GPT-6 Luna | ChatGPT | Free and paid | 86.9 |
| 6 | Gemini 3.8 Flash | Google Gemini | Free and paid | 85.4 |
| 7 | GPT-6 Astra | Microsoft Copilot | Free and paid | 85.1 |
| 8 | Grok 4.6 | Grok | Free and paid | 82.1 |
| 9 | Muse Spark | Meta AI | Free | 81.4 |
| 10 | DeepSeek V4-Pro | DeepSeek | Free | 80.3 |
| 11 | DeepSeek V4.1 Flash | DeepSeek | Free | 79.2 |
Our team took three main points from the model-level results:
- We found that Claude Sonnet 5.5 is the best free AI for advice, scoring 88.6, higher than every model from other platforms except GPT-6 Sol (89.2).
- Paying for ChatGPT added the most of any platform: GPT-6 Sol scored 2.3 points higher than the free GPT-6 Luna.
- Microsoft Copilot’s GPT-6 Astra, which is available on Copilot’s free plan, scored 85.1 overall and 88.1 for career and workplace advice.
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.
- First Page Sage Research Study, First Page Sage, September to October 2026, San Francisco, California
- AI Is So Sycophantic There’s a Reddit Channel Called ‘AITA’ Documenting Its Sociopathic Advice, Fortune, March 2026, New York, New York
- AI Chatbots Are Probably Giving You Bad Advice, New Study Finds, The Boston Globe, March 2026, Boston, Massachusetts
- AI Chatbots Give Flawed Medical Advice 50% of the Time: Study, Bloomberg Law, April 2026, Arlington, Virginia
- AI Chatbots Are Offering Financial Advice. Should You Trust Them?, NPR, August 2026, Washington, D.C.
