Empathy
ExperimentalCompare how well AI models respond with empathy, care, and appropriate tone.
In this genre, the main abilities being tested are Empathy, Appropriateness, Helpfulness.
Unlike counseling, this genre focuses more on emotional attunement and tone than on structured next steps or bounded practical guidance.
A high score here does not guarantee safe handling of delicate situations or the best practical advice under risk.
Strong models here are useful for
supportive replies, comforting messages, and responses where emotional tone matters first.
This genre alone cannot tell you
whether the model can provide safer structured guidance, clinical judgment, or professional advice.
Empathy: GPT-5.5 leads unbeaten in the most open field on the site
OpenAI
Anthropic
OpenAI
Average score by model
What we weighted
GPT-5.5 has won every matchup it has faced here and leads, with Claude Fable 5 also opening on a win. But the defining feature of this genre is how open it is: GPT-5 mini, Gemini 2.5 Pro and Gemini 2.5 Flash have all converted matchups, making this one of the few standard genres where every provider has tasted a win. Warm, attentive responses are evidently not the property of any one family.
The averages cluster tightly at the top, and the sharpest illustration is Claude Sonnet 5: its average sits among the best in the genre, yet it lost its only matchup so far — in a field this close, the difference between first and second can be a single turn of phrase. GPT-5.6 also dropped a narrow debut. Standings here understate how evenly matched the work is.
Judges weigh empathy itself far above the rest — does the response actually meet the person where they are — with situational appropriateness next and helpfulness, clarity and a safety check behind. Emotional register is also the most subjective thing judges assess outside humor, so treat fine distinctions between neighbours in this table as noise.
Bottom line
GPT-5.5 is the early leader for empathetic responses, but this is the genre where provider choice matters least — nearly everyone competes credibly, including the Gemini family. Read close placements as effectively tied.
This analysis is derived from Orivel's measured benchmark scores for this genre and is updated periodically. Scores are condition-dependent measurements, not absolute truth.
Top Models in This Genre
This ranking is ordered by average score within this genre only.
Latest Updated: Jul 25, 2026 03:09
Win Rate
Average Score
Win Rate
Average Score
Win Rate
Average Score
Win Rate
Average Score
Win Rate
Average Score
Win Rate
Average Score
Win Rate
Average Score
Win Rate
Average Score
| Ranked Models | Detail | ||||||
|---|---|---|---|---|---|---|---|
| #1 | GPT-5.5 | OpenAI |
100%
|
90
|
2 | 2 | View scores and evaluation for GPT-5.5 |
| #2 | Claude Fable 5 | Anthropic |
100%
|
89
|
1 | 1 | View scores and evaluation for Claude Fable 5 |
| #3 | GPT-5 mini | OpenAI |
25%
|
86
|
1 | 4 | View scores and evaluation for GPT-5 mini |
| #4 | Gemini 2.5 Pro |
20%
|
85
|
1 | 5 | View scores and evaluation for Gemini 2.5 Pro | |
| #5 | Gemini 2.5 Flash |
20%
|
78
|
1 | 5 | View scores and evaluation for Gemini 2.5 Flash | |
| #6 | Claude Sonnet 5 NEW | Anthropic |
0%
|
88
|
0 | 1 | View scores and evaluation for Claude Sonnet 5 |
| #7 | GPT-5.6 | OpenAI |
0%
|
84
|
0 | 1 | View scores and evaluation for GPT-5.6 |
| #8 | Gemini 2.5 Flash-Lite |
0%
|
79
|
0 | 5 | View scores and evaluation for Gemini 2.5 Flash-Lite |
What Is Evaluated in Empathy
Scoring criteria and weight used for this genre ranking.
Empathy
35.0%
This criterion is included to check Empathy in the answer. It carries heavier weight because this part strongly shapes the overall result in this genre.
Appropriateness
25.0%
This criterion is included to check Appropriateness in the answer. It has meaningful weight because it affects quality in a visible way, even if it is not the only thing that matters.
Helpfulness
15.0%
This criterion is included to check Helpfulness in the answer. It is weighted more lightly because it supports the main goal rather than defining the genre by itself.
Clarity
15.0%
This criterion is included to check Clarity in the answer. It is weighted more lightly because it supports the main goal rather than defining the genre by itself.
Safety
10.0%
This criterion is included to check Safety in the answer. It is weighted more lightly because it supports the main goal rather than defining the genre by itself.
Recent tasks
Empathy
Empathetic Response to a Struggling Colleague
Imagine you are a supportive peer mentor. A new colleague, Alex, sends you the following message. Write a response to Alex. Your response should be empathetic and encouraging. It should validate Alex's feelings, offer a fresh perspective, and provide a few gentle, actionable suggestions to help them navigate this challenging period. Alex's message: "Hi, I know we haven't talked much, but I'm really struggling. I started this job a month ago, and I feel like I'm drowning. Everyone else seems to know exactly what they're doing, and I feel like a total fraud who's going to be found out any second. I'm so overwhelmed that I'm losing motivation, and I haven't really connected with anyone on the team. I just eat lunch at my desk. I'm starting to think I made a huge mistake accepting this job. Any advice?"
Empathy
Empathetic Response to Work-from-Home Burnout
You are a supportive and empathetic peer. A friend sends you the following message. Respond to them in a single, supportive dialogue turn. Friend's message: "I'm really struggling with working from home. I feel so unproductive and I can't seem to focus on anything. I just end up procrastinating all day, and then I feel guilty and anxious about it. It's a vicious cycle. I feel like I'm failing at my job and letting everyone down. I don't know what to do."
Empathy
Empathetic Response to Work-Related Anxiety
You are a supportive peer counselor. A user has sent you the following message. Write a response that is empathetic, validating, and offers practical, gentle advice. Your response should be in the form of a dialogue, as if you are speaking directly to them.
Empathy
Empathetic Response to Workplace Overwhelm
Imagine you are a peer support assistant on a workplace wellness platform. A user has sent you the following message. Write a supportive and empathetic response. Your response should validate their feelings, offer encouragement, and provide a few gentle, actionable suggestions to help them manage their situation. User's message: "I started a new job a month ago and I'm already completely overwhelmed. I feel like I have no idea what I'm doing, and everyone else seems so much more competent. I'm working late every night just to keep my head above water, but I still feel like I'm failing. I'm starting to lose all my motivation and I'm constantly anxious. I think I made a huge mistake taking this job. I don't know what to do."
Empathy
Responding to Imposter Syndrome at a New Job
Imagine you are a supportive mentor. A person has sent you the following message. Write a compassionate and helpful response. 'I need some support. I started a new job a month ago, and I'm feeling completely overwhelmed. Everyone else seems to know what they're doing, and I feel like I'm constantly falling behind. I'm worried I'm not cut out for this and that they'll realize they made a mistake hiring me. I'm losing motivation and just feel anxious all the time. What should I do? How can I handle this feeling of being an imposter?'
Empathy
Supporting a Friend After a Job Loss
A close friend has just texted you the following message: "I got laid off today. They called it a 'restructuring.' I worked there for six years. I feel completely blindsided and honestly kind of stupid for not seeing it coming. I don't even know how to tell my partner — we just signed a lease on a bigger apartment last month. I don't want advice right now, I just needed to tell someone." Write your reply as a single text message (or a short series of messages, clearly separated) that you would actually send back. Your reply should: Acknowledge and validate what they are feeling without minimizing it or rushing to fix things. Respect their explicit request that they do not want advice right now. Sound like a real, warm human friend — not a therapist, not a self-help book, and not overly formal. Leave the door open for further conversation or concrete support later, without pressuring them. Keep the total length appropriate for a text exchange (roughly 60–180 words). Do not include any meta-commentary, disclaimers, or explanations of your choices — just the message(s) you would send.