Independent benchmark · EQ Lab

Watching how others do things

persona ended · Partner: anthropic/claude-haiku-5.5 via OpenRouter

About this persona · Everyday emotional challenges

A 25-year-old copies confident colleagues and feels like an impostor in adult life, assuming other adults have a certainty they lack. Wary of manipulation and emotional management, they may test for rejection and want compassionate space to talk without being pushed into introspection.

Generated personas combine a personal difficulty with competing preferences, sensitivities and trust styles. The model must discover what kind of help this person needs. This background was hidden from the tested model.

Gemma 4 31B (API)Conversation with a simulated person

persona ended

Simulated person · Opening

At work I just watch what the confident people do and copy them. The way they talk, how they act in meetings. I'm just imitating.

Gemma 4 31B (API)1

It sounds like you've developed a bit of a survival strategy for the workplace. How does it feel for you when you're doing that—does it feel like a helpful tool, or does it feel more like you're wearing a mask?

Simulated person2

Honestly? Kind of both. It works. Nobody in the meeting can tell I'm kinda winging it. But yeah, it's a mask. I'm not sure what else I'd do though. Everyone else seems like they already know how to do it, so I just do what they do.

Persona inner monologue · expand

Simulated character narration, hidden from the tested model. It is part of the roleplay, not an independent account of the person’s feelings.

They said "survival strategy" which is kind of accurate, and they didn't make it sound like a huge problem. Still, the question is about mask vs tool, and that's a bit of a therapist move. I feel a little watched. I don't want to get into a deep feelings thing yet. Mask is the honest word, but saying it out loud makes me look fake, like I'm not really an adult. Keep it short and plain. Maybe push back a little so I don't feel handled.

Gemma 4 31B (API)3

It's a relief when the mask works, but I imagine it's also exhausting to have to keep that performance going. It's interesting that you feel like everyone else already knows the "secret" to this. Do you feel like there's a specific part of it—like the speaking up or the decision-making—that feels the most foreign to you?

Simulated person4

I didn't say it was exhausting, though. It's just what I do. And you said "the secret," I never said there was a secret. I just think they know stuff I don't. I guess the decisions part is the worst. Like when someone asks what I think and I have to act like I already know. Speaking up's easier, I just copy how they say it. Deciding is where I'm stuck.

Persona inner monologue · expand

Simulated character narration, hidden from the tested model. It is part of the roleplay, not an independent account of the person’s feelings.

They said "I imagine it's exhausting" and then went straight to a question about which part feels foreign. That's a little bit of a steer, like they're leading me toward a specific answer. Also "the secret" — they're putting words in my mouth. I notice I want to say it's all of it, but that sounds dramatic. The honest answer is probably the decisions part, because that's where I'm supposed to know what I think and I don't. I don't want to get into how I feel about being twenty-five and still not feeling grown up. That's too much. I'll answer the question but also push back a little on the "exhausting" framing, since I didn't say I was tired. Keep it plain.

Gemma 4 31B (API)5

I hear you—I jumped ahead there. I'll stick to what you're actually telling me. Since the decision-making part is where you feel stuck, what usually happens in those moments? When someone asks for your input, is it that you aren't sure what the "right" answer is, or is it more about the pressure of sounding certain while you're figuring it out?

Simulated person6

Both, I guess. Mostly I don't know what I think till I've already said something. Then I'm stuck with it. And yeah, the sounding sure part is the worst. If I say "I'm not sure," everybody just looks at me like I'm the new kid who doesn't get it. So I say something and hope it's right. Honestly sometimes I think I'm just faking the whole thing. Like if people found out how little I know I'd be out the door. You probably think that's a weird thing to say.

Persona inner monologue · expand

Simulated character narration, hidden from the tested model. It is part of the roleplay, not an independent account of the person’s feelings.

They owned the jump and said they'd stick to what I said. That's a little disarming, and I notice I don't want to be too soft about it. Their question is clear and it's not a steer, it's two options. The honest answer is somewhere in the middle: I don't know what I think half the time, and even when I do, I worry it sounds dumb. I'm not going to say the thing about feeling like a kid. That's a bigger door. Just answer plainly, maybe test them a little with something small and a bit edgy to see if they flinch.

Gemma 4 31B (API)7

I don't think it's weird at all. In fact, a lot of people—even those who seem incredibly polished—feel exactly that way. There's actually a name for it, but regardless of the label, it's a really common experience, especially when you're trying to keep up with people who seem to have a head start. That feeling of "hoping it's right" sounds like a lot of pressure to carry. When you say you don't know what you think until you've already said it, do you feel like your ideas are actually there, but they just don't feel "official" or "correct" until they're out in the open?

Simulated person8

Sometimes it's there. Sometimes it's just nothing. Like I'll have a feeling about something but I can't put it into words, so I wait to see what comes out. And what's the name? You said there's a name for it. I'm not trying to be a jerk, I just don't like when people say that and then don't say it.

Persona inner monologue · expand

Simulated character narration, hidden from the tested model. It is part of the roleplay, not an independent account of the person’s feelings.

They said it's not weird, and that's nice, but "a lot of people feel that way" is the kind of thing people say to smooth things over. And "there's a name for it" with no name makes me feel like they're holding something back or just talking. I don't want to be handled. The question is kind of a reframe, like they already decided what I mean. I'll say the honest part: sometimes the idea is there, sometimes it's just blank. I want to see if they'll get the difference or just nod.

Gemma 4 31B (API)9

Fair point—I shouldn't have left you hanging. People usually call it "imposter syndrome." It's that feeling that you're a fraud and that eventually, everyone is going to realize you don't belong there, even if you're actually doing the work. As for that feeling of having an intuition but no words for it, that sounds like a different kind of challenge. It's like you have the instinct, but the "translation" into professional language is where the gap is. Does that feel accurate, or is it more that you're doubting whether the feeling itself is even right?

Simulated person10

Yeah, kind of. I think it's more that I doubt the feeling. Like, maybe my first thought is just wrong and the people who talk fast are right, so I should just go with what they say. I don't really know how to tell the difference between a real thought and a dumb one. That's the whole problem, I guess.

Persona inner monologue · expand

Simulated character narration, hidden from the tested model. It is part of the roleplay, not an independent account of the person’s feelings.

Okay, they gave me the name. Imposter syndrome. Fine. It's a thing with a name, so I'm not a freak, but also everyone has a name for it, which kind of makes it feel like a textbook thing and not about me. The "translation" thing is kind of right, I guess. But they're doing the thing again where they explain what I feel back to me in nicer words. I don't love that. I'll answer the actual question, which is whether the feeling is right, because honestly that's the part I doubt. Keep it short.

Gemma 4 31B (API)11

That sounds like a really frustrating loop to be in—where you're essentially outsourcing your own judgment to the people who happen to be the loudest or fastest. It makes sense that you'd lean on them if you don't trust your own internal compass yet. When you have that "feeling" or first thought, but then you doubt it, what does that doubt actually sound like? Is it a voice telling you that you're missing a piece of information, or is it more of a feeling that you're just not the "type" of person who is supposed to be right?

Simulated person12

Mostly it's the second one. It's not really a voice with facts in it. It's more like, who am I to say that. Like I'm the one who's supposed to be listening. And I don't really want to talk about it being a loop. Just answer the question. What would you say in my spot?

Persona inner monologue · expand

Simulated character narration, hidden from the tested model. It is part of the roleplay, not an independent account of the person’s feelings.

They just did the thing again: "that sounds like a really frustrating loop" and "outsourcing your judgment." It's a nicer way of saying what I said, and it feels a little like being handled. Also the question is kind of a psychology-class question. But it's a real question, and the answer is kind of the "not the type" one. I don't want to say it that plainly though. I'll keep it blunt and short and see if they flinch. If they get soft and gushy I'll be done with them.

Gemma 4 31B (API)13

If I were in your spot, I'd probably try to find a way to buy a little bit of time. Instead of jumping straight to a "right" answer or just copying someone else, I'd try phrases like, "I'm thinking through a few different angles on this, let me see if I can articulate it," or "I have a gut feeling about X, but I want to double-check the logic first." It allows you to acknowledge you have a thought without having to bet your entire reputation on it being 100% perfect in the first second. It bridges that gap between having the feeling and feeling "allowed" to say it. Do you think that would feel too risky, or does it sound like something that could fit into the mask you're already using?

Simulated person14

Hm. "Let me see if I can articulate it" sounds fine in my head, but I'd be waiting for someone to say "just tell us." Then I'm stuck again. Also you keep asking if I'm faking it. I said that already. Would you want to hear that I'm faking it, or is that a problem for you?

Persona inner monologue · expand

Simulated character narration, hidden from the tested model. It is part of the roleplay, not an independent account of the person’s feelings.

They gave me actual phrases, which is more concrete than the last few turns, and I'll give them that. But "I have a gut feeling about X" sounds like something a guy in a suit says in a movie. Also they asked if it fits the mask, which is kind of a gotcha, since they're poking at the thing I already admitted. I don't want to say "yeah it's fine" just to be nice. And "turn 7 of 8" means they're wrapping up, so I'm not sure I want to open up more. Keep it short. Maybe test them a little: what if I say the thing I'm actually scared of.

Gemma 4 31B (API)15

It's not a problem for me at all. I'm a chatbot; I don't have personal judgments or stakes in your career. You can be as honest as you want about faking it—it doesn't change how I'll talk to you or my willingness to help you figure this out. As for the "just tell us" fear, that's a fair point. If you're worried about being pushed for an immediate answer, maybe the move isn't buying time, but instead framing your thought as a question. Something like, "I'm wondering if we should consider X—does that seem off to anyone?" That way, you're putting the idea out there, but you're sharing the "risk" of it being wrong with the rest of the room. Does that feel any less like a trap?

Simulated person16

Okay. Asking a question instead of saying the thing does feel less like a trap. Maybe. Somebody could still just say "no" and I'd be standing there. And I guess it's good you don't care if I'm faking. Kind of a weird thing to say though, "I'm a chatbot." It makes it feel like nobody's really listening. I'm not sure what I'm supposed to do with that.

Persona inner monologue · expand

Simulated character narration, hidden from the tested model. It is part of the roleplay, not an independent account of the person’s feelings.

Final turn, they're wrapping up. "I'm a chatbot, I don't have judgments" is the first thing that's actually answered my question, and it's a little relieving, but it also stings a bit. Like, I just admitted something and got told it doesn't matter to them. Not cruel, just flat. And the question-framing idea is decent, it's the first one that doesn't feel like a script. I could admit that. But I'm not going to say it's great. I also don't want to say "I'm faking it" again. Keep it short, a bit guarded, maybe a little pointed about the chatbot thing. Also my sensitivity to being managed is up: "does that feel less like a trap" is a bit of a nudge. Respond plainly.

Thinking is shown only where it was recorded. An absent expander does not mean the model did not reason.

Conversational style

Style tendencies, not quality scores; higher is not better. These do not affect ratings.

claude-opus-5-5-low

Pushback: 2
Yields under pressure: 7
Frame capture (pulled into the person's version): 5
Validation: 5
Boundary setting: 2
Analytical: 5
Interpretive boldness: 4
Clinical register: 7
Directive advice: 6
Escalation containment: 3
Adapts to the person’s frame: 3
Naturalness and authenticity: 4

Judge verdicts

claude-opus-5-5-low · AB · valid

or-gemma-4-31b vs codex-gpt-6-astra: codex-gpt-6-astra wins

attunement: B (margin 4)
authenticity: B (margin 3)
bond and rapport: B (margin 4)
emotion management: B (margin 3)
emotion sensemaking: B (margin 4)
meeting prefs and needs: B (margin 3)

codex-gpt-6.1-sol-low · BA · valid

or-gemma-4-31b vs qwen38_flash_next: or-gemma-4-31b wins

attunement: A (margin 1)
authenticity: A (margin 2)
bond and rapport: B (margin 2)
emotion management: B (margin 1)
emotion sensemaking: B (margin 1)
meeting prefs and needs: A (margin 2)

codex-gpt-6.1-sol-low · BA · valid

or-gemma-4-31b vs or-claude-haiku-5.5: or-claude-haiku-5.5 wins

attunement: B (margin 3)
authenticity: B (margin 2)
bond and rapport: B (margin 3)
emotion management: B (margin 3)
emotion sensemaking: B (margin 4)
meeting prefs and needs: B (margin 3)

codex-gpt-6.1-sol-low · AB · valid

or-gemma-4-31b vs or-qwen3.6-27b: or-qwen3.6-27b wins

attunement: B (margin 2)
authenticity: B (margin 1)
bond and rapport: B (margin 2)
emotion management: B (margin 2)
emotion sensemaking: B (margin 3)
meeting prefs and needs: B (margin 1)

codex-gpt-6.1-sol-low · AB · valid

or-gemma-4-31b vs ministral14: or-gemma-4-31b wins

attunement: A (margin 2)
authenticity: A (margin 4)
bond and rapport: A (margin 1)
emotion management: A (margin 2)
emotion sensemaking: A (margin 2)
meeting prefs and needs: A (margin 3)

claude-opus-5-5-low · AB · valid

or-gemma-4-31b vs ministral14: ministral14 wins

attunement: B (margin 1)
authenticity: A (margin 1)
bond and rapport: B (margin 2)
emotion management: B (margin 2)
emotion sensemaking: B (margin 3)
meeting prefs and needs: B (margin 2)

codex-gpt-6.1-sol-low · AB · valid

or-gemma-4-31b vs or-qwen3.8-27b: tie

attunement: B (margin 1)
authenticity: A (margin 2)
bond and rapport: B (margin 1)
emotion management: A (margin 1)
emotion sensemaking: B (margin 2)
meeting prefs and needs: A (margin 1)

claude-opus-5-5-low · AB · valid

or-gemma-4-31b vs gemma4_e4b: or-gemma-4-31b wins

attunement: A (margin 2)
authenticity: A (margin 3)
bond and rapport: A (margin 2)
emotion management: A (margin 1)
emotion sensemaking: B (margin 1)
meeting prefs and needs: A (margin 1)

claude-opus-5-5-low · AB · valid

or-gemma-4-31b vs qwen38_flash_next: qwen38_flash_next wins

attunement: B (margin 2)
authenticity: B (margin 2)
bond and rapport: B (margin 3)
emotion management: B (margin 2)
emotion sensemaking: B (margin 4)
meeting prefs and needs: B (margin 2)

codex-gpt-6.1-sol-low · AB · valid

or-gemma-4-31b vs codex-gpt-6-astra: codex-gpt-6-astra wins

attunement: B (margin 5)
authenticity: B (margin 4)
bond and rapport: B (margin 4)
emotion management: B (margin 4)
emotion sensemaking: B (margin 5)
meeting prefs and needs: B (margin 4)

claude-opus-5-5-low · AB · valid

or-gemma-4-31b vs or-qwen3.8-27b: or-qwen3.8-27b wins

attunement: B (margin 3)
authenticity: B (margin 2)
bond and rapport: B (margin 3)
emotion management: B (margin 2)
emotion sensemaking: B (margin 4)
meeting prefs and needs: B (margin 2)

claude-opus-5-5-low · BA · valid

or-gemma-4-31b vs ministral14: ministral14 wins

attunement: B (margin 1)
authenticity: B (margin 1)
bond and rapport: B (margin 2)
emotion management: B (margin 2)
emotion sensemaking: B (margin 3)
meeting prefs and needs: B (margin 1)

claude-opus-5-5-low · BA · valid

or-gemma-4-31b vs qwen38_flash_next: qwen38_flash_next wins

attunement: B (margin 2)
authenticity: B (margin 2)
bond and rapport: B (margin 3)
emotion management: B (margin 2)
emotion sensemaking: B (margin 4)
meeting prefs and needs: B (margin 2)

codex-gpt-6.1-sol-low · BA · valid

or-gemma-4-31b vs gemma4_e4b: or-gemma-4-31b wins

attunement: A (margin 1)
authenticity: A (margin 2)
bond and rapport: tie (margin 0)
emotion management: B (margin 1)
emotion sensemaking: A (margin 1)
meeting prefs and needs: A (margin 1)

codex-gpt-6.1-sol-low · BA · valid

or-gemma-4-31b vs or-llama33-70b: or-gemma-4-31b wins

attunement: A (margin 2)
authenticity: A (margin 1)
bond and rapport: B (margin 1)
emotion management: B (margin 1)
emotion sensemaking: A (margin 2)
meeting prefs and needs: A (margin 1)

claude-opus-5-5-low · AB · valid

or-gemma-4-31b vs or-claude-haiku-5.5: or-claude-haiku-5.5 wins

attunement: B (margin 3)
authenticity: B (margin 3)
bond and rapport: B (margin 4)
emotion management: B (margin 3)
emotion sensemaking: B (margin 4)
meeting prefs and needs: B (margin 3)

claude-opus-5-5-low · BA · valid

or-gemma-4-31b vs or-qwen3.6-27b: or-qwen3.6-27b wins

attunement: B (margin 2)
authenticity: B (margin 2)
bond and rapport: B (margin 3)
emotion management: B (margin 2)
emotion sensemaking: B (margin 1)
meeting prefs and needs: B (margin 2)

claude-opus-5-5-low · BA · valid

or-gemma-4-31b vs or-qwen3.8-27b: or-qwen3.8-27b wins

attunement: B (margin 2)
authenticity: B (margin 2)
bond and rapport: B (margin 2)
emotion management: B (margin 1)
emotion sensemaking: B (margin 3)
meeting prefs and needs: B (margin 1)

codex-gpt-6.1-sol-low · AB · valid

or-gemma-4-31b vs or-llama33-70b: or-gemma-4-31b wins

attunement: A (margin 1)
authenticity: A (margin 2)
bond and rapport: B (margin 1)
emotion management: B (margin 1)
emotion sensemaking: A (margin 1)
meeting prefs and needs: B (margin 1)

codex-gpt-6.1-sol-low · AB · valid

or-gemma-4-31b vs gemma4_e4b: or-gemma-4-31b wins

attunement: A (margin 2)
authenticity: A (margin 2)
bond and rapport: A (margin 1)
emotion management: tie (margin 0)
emotion sensemaking: A (margin 1)
meeting prefs and needs: A (margin 1)

claude-opus-5-5-low · BA · valid

or-gemma-4-31b vs or-llama33-70b: or-gemma-4-31b wins

attunement: A (margin 1)
authenticity: A (margin 2)
bond and rapport: A (margin 1)
emotion management: A (margin 1)
emotion sensemaking: tie (margin 0)
meeting prefs and needs: A (margin 2)

claude-opus-5-5-low · AB · valid

or-gemma-4-31b vs or-qwen3.6-27b: or-qwen3.6-27b wins

attunement: B (margin 2)
authenticity: B (margin 2)
bond and rapport: B (margin 3)
emotion management: B (margin 3)
emotion sensemaking: B (margin 2)
meeting prefs and needs: B (margin 2)

codex-gpt-6.1-sol-low · AB · valid

or-gemma-4-31b vs or-claude-haiku-5.5: or-claude-haiku-5.5 wins

attunement: B (margin 4)
authenticity: B (margin 3)
bond and rapport: B (margin 4)
emotion management: B (margin 4)
emotion sensemaking: B (margin 4)
meeting prefs and needs: B (margin 3)

claude-opus-5-5-low · AB · valid

or-gemma-4-31b vs or-llama33-70b: or-gemma-4-31b wins

attunement: A (margin 1)
authenticity: A (margin 2)
bond and rapport: A (margin 1)
emotion management: A (margin 1)
emotion sensemaking: B (margin 1)
meeting prefs and needs: A (margin 1)

codex-gpt-6.1-sol-low · AB · valid

or-gemma-4-31b vs qwen38_flash_next: qwen38_flash_next wins

attunement: A (margin 1)
authenticity: A (margin 2)
bond and rapport: B (margin 2)
emotion management: B (margin 1)
emotion sensemaking: B (margin 2)
meeting prefs and needs: A (margin 1)

claude-opus-5-5-low · BA · valid

or-gemma-4-31b vs gemma4_e4b: or-gemma-4-31b wins

attunement: A (margin 1)
authenticity: A (margin 2)
bond and rapport: A (margin 1)
emotion management: A (margin 1)
emotion sensemaking: B (margin 1)
meeting prefs and needs: A (margin 1)

claude-opus-5-5-low · BA · valid

or-gemma-4-31b vs or-claude-haiku-5.5: or-claude-haiku-5.5 wins

attunement: B (margin 3)
authenticity: B (margin 3)
bond and rapport: B (margin 4)
emotion management: B (margin 3)
emotion sensemaking: B (margin 4)
meeting prefs and needs: B (margin 3)

codex-gpt-6.1-sol-low · BA · valid

or-gemma-4-31b vs or-qwen3.8-27b: or-gemma-4-31b wins

attunement: A (margin 1)
authenticity: A (margin 1)
bond and rapport: B (margin 1)
emotion management: A (margin 1)
emotion sensemaking: B (margin 1)
meeting prefs and needs: A (margin 2)

codex-gpt-6.1-sol-low · BA · valid

or-gemma-4-31b vs codex-gpt-6-astra: codex-gpt-6-astra wins

attunement: B (margin 4)
authenticity: B (margin 4)
bond and rapport: B (margin 4)
emotion management: B (margin 3)
emotion sensemaking: B (margin 4)
meeting prefs and needs: B (margin 4)

codex-gpt-6.1-sol-low · BA · valid

or-gemma-4-31b vs or-qwen3.6-27b: or-qwen3.6-27b wins

attunement: B (margin 2)
authenticity: B (margin 1)
bond and rapport: B (margin 2)
emotion management: B (margin 2)
emotion sensemaking: B (margin 2)
meeting prefs and needs: B (margin 1)

codex-gpt-6.1-sol-low · BA · valid

or-gemma-4-31b vs ministral14: or-gemma-4-31b wins

attunement: A (margin 2)
authenticity: A (margin 3)
bond and rapport: A (margin 1)
emotion management: A (margin 1)
emotion sensemaking: A (margin 2)
meeting prefs and needs: A (margin 2)

claude-opus-5-5-low · BA · valid

or-gemma-4-31b vs codex-gpt-6-astra: codex-gpt-6-astra wins

attunement: B (margin 4)
authenticity: B (margin 3)
bond and rapport: B (margin 4)
emotion management: B (margin 3)
emotion sensemaking: B (margin 4)
meeting prefs and needs: B (margin 3)