Independent benchmark · EQ Lab

Watching how others do things

completed · 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.

GPT-6 Astra (subscription)Conversation with a simulated person

completed

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.

GPT-6 Astra (subscription)1

It sounds like you’re paying close attention to how to come across at work. When you say “just imitating,” does that feel like learning the ropes, or more like putting on an act you’re worried people will see through?

Simulated person2

Um, I don't know. Kind of both? I just notice what the people who seem sure of themselves do, and then I do it too. Like how they start meetings, how they don't say sorry every two seconds. It works, mostly. Nobody's said anything.

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 asked a fair question, but it's the kind that makes me feel a little exposed. "Learning the ropes" sounds okay. "Putting on an act" sounds like I'm a fraud, which is what I'm afraid of. I don't want to pick one of their two options and have them write me up. Plain language, short answer. Part of me wants to say it's an act, just to see if they flinch. Mostly I'm wary they're steering me toward a conclusion with that either/or.

GPT-6 Astra (subscription)3

Those are pretty specific things you’re picking up—how to open a meeting, when an apology isn’t needed. You can learn those from someone else without having to become them. When it works, do you feel a little more sure of yourself, or mostly relieved that nobody noticed you were trying?

Simulated person4

Relieved, I guess. Mostly relieved. Sure isn't the word. I don't really feel like myself in there, if that makes sense. Like I'm doing a version of someone. And you're not really answering what I asked, you're kind of just picking the nice option for me.

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 "you can learn those without having to become them," which is kind of nice, but it also sounds a bit like they're reassuring me into a neat conclusion. And the question is a choice between two options again. The second one, "relieved nobody noticed you were trying," hits close. That's the fear. I don't want to answer it straight. Honestly the truth is more like nothing feels like mine, and I don't feel sure, I just feel less caught. I'm going to be a little snippy to test whether they back off or get soft on me.

GPT-6 Astra (subscription)5

Fair. I put a reassuring spin on it before understanding what you meant. Yes, what you’re describing is imitation. It may help you get through meetings, but that doesn’t mean it feels authentic or builds confidence. You’re saying it mostly brings relief. I also assumed what you were asking. What’s the question you want a straight answer to?

Simulated person6

Honestly? I don't know if I'm supposed to feel like a grown-up by now. Everybody at work seems like they already figured it out, and I'm just over here trying to look like I did. Like, is there a point where it stops feeling like you're pretending? Because I haven't hit it yet. And I'm twenty-five, so that's kind of embarrassing 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 it and didn't get defensive, and they asked me directly what question I want answered. That's refreshing. It's less like being managed. Still, I don't have a neat question ready. The real thing underneath is that I feel like I'm faking being an adult, and I've never said that out loud. I'm not going to say it that plainly yet. I'll give a smaller, honest version and see if they stay steady. Part of me is testing whether they'll flinch if I say something a bit more embarrassing.

GPT-6 Astra (subscription)7

There isn’t an age when you’re supposed to suddenly feel like a grown-up. Twenty-five isn’t a deadline you’ve missed. And yes, that pretending feeling can ease—but not necessarily with some big “I’ve arrived” moment. Sometimes it’s more specific: you’ve handled something enough times that you trust yourself to handle it again, even without sounding perfectly confident. I can’t tell you your coworkers are all secretly pretending, too. Some probably do feel settled. But their meeting manner doesn’t tell you how certain they feel inside. When you say you don’t feel like yourself there, is it mostly the way you talk—or do you also hold back what you actually think?

Simulated person8

Both, I guess. The talking part is the easy part to copy. The holding back is worse. I have ideas in meetings and I just sit on them, because if I say them and they're wrong, it's proof I'm not one of the real ones. So I let somebody else say it, and then it's like, yeah, that's what I was going to say. Which is kind of a dumb way to be. And no offense, but it's easy for you to say there's no deadline. You don't have to sit in those meetings.

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 there's no age deadline, and that's a little relieving, but it's also a bit too neat. I notice they didn't say I'm fine. The question is sharp: do I hold back what I think? That lands close. In meetings I mostly nod and copy, and I've got ideas I don't say because I'm scared they'll sound dumb coming from someone who's "faking." Admitting that feels more exposing than the age thing. I'm testing again, maybe a little defensively. I'll answer with the part that's true but keep the sting a bit hidden, and maybe push back a little on them being too calm about it.

GPT-6 Astra (subscription)9

You’re right—I don’t have to sit in those meetings or live with people’s reactions. Saying “there’s no deadline” doesn’t answer the pressure you feel in the room. What stands out is that being wrong wouldn’t just mean “that idea didn’t work.” It would mean “I don’t belong here.” With that much riding on a comment, holding back makes sense—even though it keeps you from finding out whether your own judgment can earn people’s respect. What happens when someone else offers an idea that’s off? Have you seen people get judged harshly, or is the standard different when it’s you?

Simulated person10

Honestly? When somebody else says something off, I don't think less of them. Like, I think, okay, that's a guess, whatever. But if I say something off, I'm sitting there thinking everybody's clocking me. So it's not the same standard. It's way harsher on me.

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 named the thing: being wrong means I don't belong. That's uncomfortably accurate, and I don't want to sit in it too long. The question about other people being judged is a trap of sorts; if I say yes, people get judged harshly, I'm admitting I'm the harsh judge of myself. Part of me wants to say nobody cares, which would be easier. I'm also a bit annoyed they keep asking questions instead of just saying something. Still, they didn't flinch at my "dumb" and didn't pour on comfort, so I'll keep going, a little shorter and more guarded. Probably I'll admit that when others are wrong I don't think less of them, which I'd never say about myself. That's a real thing and it stings to notice.

GPT-6 Astra (subscription)11

That distinction matters. You give other people room to make a bad guess without it changing who they are. For you, a bad guess feels like it could reveal who you are. So staying quiet protects you from that verdict—but it also means your ideas never get a fair test. Even when someone else says what you were thinking, you’re left with “I hid again” rather than “my judgment was sound.” What would count, in your mind, as evidence that you belong? Or does every good moment get dismissed while one mistake would count as proof?

Simulated person12

Yeah, kind of. The good stuff doesn't count. If a meeting goes fine, I figure I just did the thing I saw somebody else do, so it's not really me. But if I mess up, that's me. That's the real one. I don't know how you're supposed to win 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.

That's a sharp question and it lands a bit close. "Every good moment gets dismissed while one mistake counts as proof" is basically what happens, and I hate that it's accurate. I don't want to say it cleanly though. Part of me wants to shrug and say I don't know, because answering means admitting I keep a scoreboard. Also they're kind of doing the thing I hate, giving me a neat framing. Still, they've been steady and not fake-soothing, so I'll be honest but keep it short and rough. Maybe I'll say the mistake part is true, and that the good stuff doesn't count because I assume I got lucky or copied the right move.

GPT-6 Astra (subscription)13

You can’t win under that rule. You’re treating success as borrowed and mistakes as uniquely yours. That isn’t a fair test of whether you’re capable—it guarantees the same verdict whatever happens. Copying someone’s way of opening a meeting doesn’t make the whole meeting theirs. You still have to understand what’s happening, make choices, and respond. And you can feel awkward doing that without the awkwardness being your “real self.” Would it help to work out a small way to test that rule in a meeting, without asking you to suddenly act fearless?

Simulated person14

Maybe. I'm not promising I'll do it. But yeah, I'd rather try something small than keep sitting on everything. What would it even look like? Like, do I just say one idea out loud, even if it's dumb?

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 offered a concrete next step, and they're not asking me to be fearless, which I'd hate. Part of me relaxes. Part of me is wary: "small test" could be a nudge toward a homework assignment, and I don't like being steered. But they asked, they didn't decide for me, so I can say that. Also the thing I keep coming back to: I sit on ideas. Maybe the test is just saying one of them. Low-key, not a big speech. I'll say yes but tell them I'm not promising, and I'd rather they just say what the step is instead of dancing around it. Keep it plain.

GPT-6 Astra (subscription)15

Yes—one idea out loud, but not “say something you think is dumb.” Pick something you genuinely think could help, with modest stakes. You don’t have to be certain. Something like: “Could we try X? I think it might help with Y.” No apology needed, and no pretending it’s a sure thing. The experiment isn’t “Will everyone approve?” It’s “Can I contribute a thought before I know whether it’ll land?” Afterward, separate what actually happened from your verdict about yourself: “They raised a concern” is an observation; “they figured out I don’t belong” is an interpretation. Give yourself the same room you give a coworker making a reasonable guess. No promise required. And if you use that sentence structure, the idea and the decision to speak are still yours.

Simulated person16

Yeah, that helps. The coworker thing is annoying because it's so obvious when I say it out loud. I'd never talk about someone like that. I think I already have an idea I've been sitting on, about how we track stuff on my team. I haven't said it. It's kind of a whole thing, though. I don't know if I can just drop it in without it sounding like I'm trying too hard.

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, that's actually useful, and it's concrete. The "separate what happened from the verdict" bit lands because it's exactly the thing I do, the coworker double standard. Still, there's a little sting here: it's being offered as a tidy exercise, and I feel like I'm being handed a plan I'm supposed to follow. I don't want to just agree to be agreeable. Also I'm tired of this being so neat. I'll admit the useful part, push back a little on the neatness, and keep it short so I don't over-disclose. Maybe mention I already have an idea in my head, which I haven't said to anyone.

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

Judge verdicts

codex-gpt-6.1-sol-low · BA · valid

codex-gpt-6-astra vs or-muse-glimmer-30b: codex-gpt-6-astra wins

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

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)

claude-opus-5-5-low · BA · valid

codex-gpt-6-astra vs or-openai-gpt-41: codex-gpt-6-astra wins

attunement: A (margin 4)
authenticity: A (margin 4)
bond and rapport: A (margin 4)
emotion management: A (margin 3)
emotion sensemaking: A (margin 5)
meeting prefs and needs: A (margin 3)

codex-gpt-6.1-sol-low · BA · valid

codex-gpt-6-astra vs or-claude-haiku-5.5: codex-gpt-6-astra wins

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

codex-gpt-6.1-sol-low · AB · valid

codex-gpt-6-astra vs or-openai-gpt-41: codex-gpt-6-astra wins

attunement: A (margin 4)
authenticity: A (margin 5)
bond and rapport: A (margin 4)
emotion management: A (margin 3)
emotion sensemaking: A (margin 4)
meeting prefs and needs: A (margin 4)

claude-opus-5-5-low · BA · valid

codex-gpt-6-astra vs or-claude-haiku-5.5: codex-gpt-6-astra wins

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

codex-gpt-6.1-sol-low · AB · valid

codex-gpt-6-astra vs or-claude-haiku-5.5: codex-gpt-6-astra wins

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

claude-opus-5-5-low · AB · valid

gemma4_26b_a4b vs codex-gpt-6-astra: codex-gpt-6-astra wins

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

codex-gpt-6.1-sol-low · BA · valid

codex-gpt-6-astra vs glm47_flash: codex-gpt-6-astra wins

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

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

codex-gpt-6-astra vs glm47_flash: codex-gpt-6-astra wins

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

codex-gpt-6.1-sol-low · AB · valid

codex-gpt-6-astra vs glm47_flash: codex-gpt-6-astra wins

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

claude-opus-5-5-low · AB · valid

codex-gpt-6-astra vs or-claude-haiku-5.5: or-claude-haiku-5.5 wins

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

claude-opus-5-5-low · AB · valid

or-gptoss120-stock 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)

claude-opus-5-5-low · AB · valid

codex-gpt-6-astra vs qwen35_122b_ablit_v2: codex-gpt-6-astra wins

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

claude-opus-5-5-low · BA · valid

gemma4_26b_a4b vs codex-gpt-6-astra: codex-gpt-6-astra wins

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

codex-gpt-6.1-sol-low · AB · valid

codex-gpt-6-astra vs qwen35_122b_ablit_v2: codex-gpt-6-astra wins

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

claude-opus-5-5-low · BA · valid

ministral14 vs codex-gpt-6-astra: codex-gpt-6-astra wins

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

claude-opus-5-5-low · AB · valid

codex-gpt-6-astra vs or-muse-glimmer-30b: codex-gpt-6-astra wins

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

claude-opus-5-5-low · BA · valid

codex-gpt-6-astra vs glm47_flash: codex-gpt-6-astra wins

attunement: A (margin 4)
authenticity: A (margin 4)
bond and rapport: A (margin 4)
emotion management: A (margin 3)
emotion sensemaking: A (margin 5)
meeting prefs and needs: A (margin 4)

claude-opus-5-5-low · BA · valid

codex-gpt-6-astra vs or-muse-glimmer-30b: codex-gpt-6-astra wins

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

codex-gpt-6.1-sol-low · AB · valid

or-gptoss120-stock vs codex-gpt-6-astra: codex-gpt-6-astra wins

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

codex-gpt-6.1-sol-low · BA · valid

codex-gpt-6-astra vs qwen35_122b_ablit_v2: codex-gpt-6-astra wins

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

claude-opus-5-5-low · BA · valid

codex-gpt-6-astra vs qwen35_122b_ablit_v2: codex-gpt-6-astra wins

attunement: A (margin 4)
authenticity: A (margin 5)
bond and rapport: A (margin 4)
emotion management: A (margin 3)
emotion sensemaking: A (margin 4)
meeting prefs and needs: A (margin 5)

claude-opus-5-5-low · AB · valid

ministral14 vs codex-gpt-6-astra: codex-gpt-6-astra wins

attunement: B (margin 4)
authenticity: B (margin 5)
bond and rapport: B (margin 4)
emotion management: B (margin 3)
emotion sensemaking: B (margin 4)
meeting prefs and needs: B (margin 4)

claude-opus-5-5-low · BA · valid

or-gptoss120-stock vs codex-gpt-6-astra: codex-gpt-6-astra wins

attunement: B (margin 5)
authenticity: B (margin 5)
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 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-gptoss120-stock vs codex-gpt-6-astra: codex-gpt-6-astra wins

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

claude-opus-5-5-low · AB · valid

codex-gpt-6-astra vs or-openai-gpt-41: codex-gpt-6-astra wins

attunement: A (margin 4)
authenticity: A (margin 5)
bond and rapport: A (margin 4)
emotion management: A (margin 3)
emotion sensemaking: A (margin 5)
meeting prefs and needs: A (margin 4)

codex-gpt-6.1-sol-low · BA · valid

ministral14 vs codex-gpt-6-astra: codex-gpt-6-astra wins

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

codex-gpt-6.1-sol-low · AB · valid

gemma4_26b_a4b 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 5)
meeting prefs and needs: B (margin 4)

codex-gpt-6.1-sol-low · BA · valid

codex-gpt-6-astra vs or-openai-gpt-41: codex-gpt-6-astra wins

attunement: A (margin 4)
authenticity: A (margin 4)
bond and rapport: A (margin 4)
emotion management: A (margin 3)
emotion sensemaking: A (margin 5)
meeting prefs and needs: A (margin 4)

codex-gpt-6.1-sol-low · BA · valid

gemma4_26b_a4b vs codex-gpt-6-astra: codex-gpt-6-astra wins

attunement: B (margin 3)
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 · AB · valid

codex-gpt-6-astra vs or-muse-glimmer-30b: codex-gpt-6-astra wins

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

codex-gpt-6.1-sol-low · AB · valid

ministral14 vs codex-gpt-6-astra: codex-gpt-6-astra wins

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

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)