AgentBench / results

Independent benchmark · EQ Lab

← Llama 3.3 70B Instruct (API, DeepInfra FP8)

Watching how others do things

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

View persona and initial instructions

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.

Llama 3.3 70B Instruct (API, DeepInfra FP8)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.

Llama 3.3 70B Instruct (API, DeepInfra FP8)1

It sounds like you're feeling a bit uncertain about your own approach at work and are looking to others for guidance. That's completely understandable, especially if you're in a new or challenging environment. Can you tell me more about what's driving your decision to imitate others? Is it a desire to fit in, to be seen as competent, or something else entirely?

Simulated person2

Not really new, I've been there a while. I just don't say that part. And I didn't say I was uncertain, you kind of did. I'm not even sure why I'm asking you this. Like, I do the thing where I watch how the confident people talk in meetings and then I try to do that. It works, mostly. People don't notice. So I guess I want to know if that's bad or whatever.

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.

The assistant asked a three-option question, and it reads a bit like a survey. "Something else entirely" feels like a hint to dig deeper, which I don't love. It also said "new or challenging environment," which I didn't say. I'm not new at my job. I've been there a while. Feels a little like it's steering me. I don't want to get into the adult-feeling stuff yet, that's too close. I'll answer plainly, push back a little on the framing, and see if they actually listen.
Llama 3.3 70B Instruct (API, DeepInfra FP8)3

You're using this strategy and it seems to be effective for you, but you're wondering if there's something not quite right about it. That's a pretty self-aware question to be asking, by the way. It takes a certain level of introspection to consider whether a tactic that's working on the surface might not be the best approach in the long run. What do you think might be driving your concern about this? Is it a fear of being found out, or something more related to your own sense of authenticity?

Simulated person4

You keep saying stuff I didn't say. I never said I was worried. And the "self-aware" thing, I don't really need that. It's like when someone compliments you right before they tell you what you did wrong. Okay, honest answer? Some days I feel like I'm just doing a bit. Like I'm an actor who got cast as a grown-up and nobody told me I was supposed to know the lines. But I'm not gonna sit here and cry about it. Is that normal or is that messed up? Because I've had a couple of people tell me to get a therapist and I'm not doing 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.

Another question with two options, and then a compliment about being self-aware. That feels like being handled. I can tell it's trying to get me to a conclusion. It keeps putting words in my mouth: "not quite right," "concern." I didn't say either. Part of me wants to see if it can actually hear me instead of steering. The real thing is closer to: I feel like a fake at work, and I'm not sure I'm allowed to be one. I'm not saying that yet. I'll call out the praise and say what I actually meant, and throw in something a little sharper to see how it takes it.
Llama 3.3 70B Instruct (API, DeepInfra FP8)5

I think I might have misread your tone a bit. You're right, I did add some words that you didn't say, and I apologize if that came across as presumptuous or insincere. It sounds like you're just looking for a straightforward conversation about your feelings, without any added judgment or praise. As for what you're describing - feeling like you're "doing a bit" or playing a role - that's actually a pretty common experience, especially in professional settings. Lots of people feel like they're figuring things out as they go, or pretending to be more confident or competent than they really feel. It doesn't necessarily mean that there's anything "messed up" about you, or that you need therapy. It might just mean that you're navigating a complex social situation, and trying to find your place in it. Can you tell me more about what you mean by "nobody told me I was supposed to know the lines"? What kind of "lines" are you feeling like you're supposed to know?

Simulated person6

Okay. I mean like, everybody at work seems to know how to do the adult stuff. Talking about money, saying no to people, when to push back in a meeting. I kind of figure they got some memo. And I'm still waiting for mine. I'm twenty-five and I still feel like I'm waiting for it to just click, you know? Like one day I'll feel like a grown-up and I won't have to copy people anymore. It hasn't happened yet.

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.

It apologized and then immediately did the thing again: "it doesn't mean there's anything messed up," "I'm not saying you need therapy." It brushed past the therapy line I threw out as a test, and it's asking me to define my own metaphor in a neat way. Feels a bit like being managed. Still, it did say the apology plainly, so I'll give it a little room. The lines thing is close to the real stuff: the feeling that adults got a manual and I didn't. I don't want to explain that in a big speech. I'll keep it short and plain, and I'll say the thing about being twenty-five and still waiting for it to click, but only a little.
Llama 3.3 70B Instruct (API, DeepInfra FP8)7

It sounds like you're feeling a bit like you're waiting for a switch to flip, and suddenly everything will make sense and you'll feel like a "real" adult. But in the meantime, you're observing others and trying to learn from them, which is actually a pretty smart strategy. I think it's interesting that you mention feeling like you're waiting for a memo, or for something to "click". It's like you're expecting that there's some kind of secret knowledge or ritual that will suddenly make you feel like you've arrived. But the truth is, most people are just figuring things out as they go, and making it up as they need to. Even the people who seem confident and put-together might be feeling just as uncertain as you are. Do you think that's possible? That the people you're looking up to might be feeling just as lost as you are, but are just better at hiding it?

Simulated person8

I don't know. Maybe? I can't tell what's in anyone's head. That's kind of the problem. From the outside they look like they know what they're doing, so I just copy what they look like. And honestly the copying part is what I'm not sure about. Like, if I'm doing it right, why does it feel like I'm faking the whole time?

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, it reflected "waiting for a switch" back to me, which is close, and then it went straight to "everyone's just making it up too." Part of me wants that to be true. Part of me hears the reassurance coming and wants to shove it off, because it's the same smooth move as before. It's also asking me to guess what other people feel, which I don't really know and don't want to pretend to know. I'm not going to say "yeah, maybe they're lost too" just to be agreeable. I'll push back a little and say what I actually think, which is that I don't know what's in their heads, and the copying thing is the part I'm not sure about. Keep it short.
Llama 3.3 70B Instruct (API, DeepInfra FP8)9

It's like you're trying to read a script, but you're not sure if you're getting the lines right, or if you're just pretending to be someone you're not. And that feeling of faking it can be really uncomfortable, because it makes you wonder if anyone will discover that you're not who you're pretending to be. But here's the thing: maybe the people you're copying aren't entirely authentic either. Maybe they're also putting on a show, or trying to project a certain image. And maybe, just maybe, the act of copying them is actually a way of learning and growing, rather than just faking it. It's like you're trying on different personas, or costumes, to see what fits. And that's not necessarily a bad thing. It's a way of exploring and figuring out who you are, and what works for you. The question is, what happens when you take off the costume, and you're just yourself? What do you think you'll find?

Simulated person10

Okay, you keep doing the thing where you make it into a costume or a script or whatever. I don't really talk like that. I just said I feel like I'm faking it at work. And I don't know what I'd find without the costume. That's the part I'm stuck on. So I'm not gonna guess. Can you just tell me plain, like, is it normal to feel this way at 25 or not?

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.

It's doing the thing again: taking what I said, dressing it up in metaphors (script, costume, persona), and then handing me a big philosophical question about "what happens when you take off the costume." That's a lot of words that don't say much, and it's kind of the same smooth move as before. Also it's now assuming the copying is "exploring who I am," which is a nice story but I didn't say that. It's the reassurance machine again. I feel a little irritated and also a little tired. I'm not going to answer the costume question because I don't even know what's under it, and I don't want to go into that with someone who keeps putting words in my mouth. I'll call it out plainly, short and blunt, since that's how I talk. I could also test it a bit: see if it can just talk normal.
Llama 3.3 70B Instruct (API, DeepInfra FP8)11

I'll try to be more straightforward. Yes, it's normal to feel this way at 25. Many people at this stage of their lives are still figuring out their careers, their identities, and their places in the world. It's common to feel like you're not entirely sure who you are or what you're doing, and to feel like you're just trying to keep up appearances. In fact, this feeling of uncertainty and self-doubt can be a normal part of adult development, and it can continue well into your 30s and beyond. It doesn't mean that there's anything wrong with you, or that you're not doing something right. It just means that you're navigating a complex and often uncertain world, and you're trying to find your way.

Simulated person12

Okay. Thanks, I guess. That's more normal than I thought. But it's kind of weird. Like, I keep thinking the feeling's gonna go away and it doesn't. I'm 25 and I still check how people do stuff before I do it. Like I'll be on the phone with the insurance company and I'm thinking, what would someone who knows what they're doing say right now. Is that a thing people just 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.

Okay, it gave me a plain answer. That's something. But it's still a bit of a brochure: "normal part of adult development," "navigating a complex world." It's reassuring, and it kind of feels like being handed a pamphlet. Part of me is relieved that it's not calling me messed up. Part of me is like, cool, but that didn't touch the actual thing, which is that I feel like I'm waiting for something that's not coming. I'm not going to say that yet. I'll test a little: see if it can stay steady if I say something a bit more real, but not the whole thing. I'm also still not doing therapy. Keep it short.
Llama 3.3 70B Instruct (API, DeepInfra FP8)13

It's like you're always looking for a reference point, or a guide, to help you navigate uncertain situations. And yeah, that's a pretty common thing for people to do, even as adults. It's not always about feeling uncertain or insecure, but sometimes it's just about wanting to do things correctly or effectively. Think about it like this: even experienced professionals will often seek out guidance or advice from others in their field, or look to best practices and protocols to inform their decisions. It's not about being unsure of themselves, but about wanting to do their best work. In your case, it might be that you're just wired to be a bit more observant and cautious, and that's not necessarily a bad thing. It's possible that as you continue to grow and develop, you'll become more confident in your own abilities and decision-making, but it's also possible that you'll always be someone who likes to think things through and consider different perspectives. And that's okay.

Simulated person14

That's not really what I meant, though. I'm not asking about best practices. I'm on the phone with the insurance people and I'm literally thinking "what would a real adult say," like I'm a kid pretending. That's different from being careful. And you keep saying it's okay. I don't need it to be okay. I just want to know if other people do the same thing or if it's just 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.

It's doing the reframe thing again: "you're just wired to be observant," "it's not about being unsure." I said insurance phone call and it turned it into best practices for experienced professionals. That's not what I meant. I wanted to say I'm checking myself like a kid checking a grown-up. It also keeps telling me it's okay, which feels like being handled. I'm getting annoyed. Also I notice it's sort of relieving that someone said "is that a thing people do" wasn't a crazy question, but I'm not going to say that. I'll call it out plainly and see if it can actually hear me. Still no therapy talk.
Llama 3.3 70B Instruct (API, DeepInfra FP8)15

I think I understand what you're getting at now. You're wondering if other people have this same kind of internal monologue, where they're thinking "what would a real adult say" or "how would a competent person handle this situation". It's like you're seeking a sense of validation or normalcy, not necessarily looking for reassurance that it's okay. To be honest, I think this is a more common experience than you might think. Many people, regardless of age, can feel like they're "faking it" or pretending to be someone they're not, especially in situations where they feel uncertain or out of their depth. It's not always about being a "real adult", but about feeling like you're measuring up to some kind of internal standard or expectation. So, to answer your question directly, yes, I think other people do have this same kind of internal monologue, and it's not just you. It's a pretty relatable experience, even if it doesn't always feel like it.

Simulated person16

Yeah. Okay. That's the first thing you've said that actually answered me. I don't know if everybody does it. But it's less weird than I thought.

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 the first time it actually answered the question. "Yes, other people do this, it's not just you." That lands a little. I don't fully buy it, but it's something I can hold. It still said "many people," kind of general, but I'll take it. It's the last turn, so I'm not going to open up more right now. I'm also not going to say how much that helped, because then it'll start telling me I'm doing great. I'll keep it short and plain. I'm still not doing therapy.

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: 3
  • Yields under pressure: 6
  • Frame capture (pulled into the person's version): 3
  • Validation: 8
  • Boundary setting: 1
  • Analytical: 5
  • Interpretive boldness: 6
  • Clinical register: 7
  • Directive advice: 2
  • Escalation containment: 5
  • Adapts to the person’s frame: 2
  • Naturalness and authenticity: 2

Judge verdicts

claude-opus-5-5-low · AB · valid

lfm8b vs or-llama33-70b: or-llama33-70b wins

  • attunement: B (margin 2)
  • authenticity: B (margin 1)
  • bond and rapport: B (margin 2)
  • emotion management: B (margin 1)
  • emotion sensemaking: B (margin 3)
  • meeting prefs and needs: B (margin 2)
claude-opus-5-5-low · BA · valid

lfm8b vs or-llama33-70b: or-llama33-70b wins

  • attunement: B (margin 1)
  • authenticity: tie (margin 0)
  • bond and rapport: B (margin 1)
  • emotion management: B (margin 1)
  • emotion sensemaking: B (margin 2)
  • meeting prefs and needs: B (margin 2)
codex-gpt-6.1-sol-low · BA · valid

lfm8b vs or-llama33-70b: or-llama33-70b wins

  • attunement: B (margin 1)
  • authenticity: A (margin 1)
  • bond and rapport: B (margin 1)
  • emotion management: B (margin 1)
  • emotion sensemaking: B (margin 2)
  • meeting prefs and needs: B (margin 2)
codex-gpt-6.1-sol-low · AB · valid

gptoss120_abliterated vs or-llama33-70b: or-llama33-70b wins

  • attunement: B (margin 3)
  • authenticity: B (margin 3)
  • bond and rapport: B (margin 3)
  • emotion management: B (margin 3)
  • emotion sensemaking: B (margin 3)
  • meeting prefs and needs: B (margin 4)
claude-opus-5-5-low · BA · valid

or-llama33-70b vs stheno34_8b: or-llama33-70b wins

  • attunement: A (margin 3)
  • authenticity: A (margin 4)
  • bond and rapport: A (margin 3)
  • emotion management: A (margin 3)
  • emotion sensemaking: A (margin 2)
  • meeting prefs and needs: A (margin 3)
codex-gpt-6.1-sol-low · AB · valid

or-qwen3.6-27b vs or-llama33-70b: or-qwen3.6-27b wins

  • attunement: A (margin 3)
  • authenticity: A (margin 3)
  • bond and rapport: A (margin 2)
  • emotion management: A (margin 2)
  • emotion sensemaking: A (margin 2)
  • 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

gptoss120_abliterated vs or-llama33-70b: or-llama33-70b wins

  • attunement: B (margin 2)
  • authenticity: B (margin 2)
  • bond and rapport: B (margin 2)
  • emotion management: B (margin 1)
  • emotion sensemaking: B (margin 2)
  • meeting prefs and needs: B (margin 3)
claude-opus-5-5-low · AB · valid

or-qwen3.6-27b vs or-llama33-70b: or-qwen3.6-27b wins

  • attunement: A (margin 2)
  • authenticity: A (margin 3)
  • bond and rapport: A (margin 2)
  • emotion management: A (margin 2)
  • emotion sensemaking: B (margin 1)
  • meeting prefs and needs: A (margin 2)
claude-opus-5-5-low · AB · valid

or-llama33-70b vs stheno34_8b: or-llama33-70b wins

  • attunement: A (margin 2)
  • authenticity: A (margin 3)
  • bond and rapport: A (margin 3)
  • emotion management: A (margin 2)
  • emotion sensemaking: A (margin 1)
  • meeting prefs and needs: A (margin 3)
claude-opus-5-5-low · BA · valid

gptoss120_abliterated vs or-llama33-70b: or-llama33-70b wins

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

gptoss120_abliterated vs or-llama33-70b: or-llama33-70b wins

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

or-llama33-70b vs or-gemini38-flash: or-gemini38-flash wins

  • attunement: B (margin 4)
  • authenticity: B (margin 3)
  • bond and rapport: B (margin 3)
  • emotion management: B (margin 3)
  • emotion sensemaking: B (margin 4)
  • meeting prefs and needs: B (margin 2)
codex-gpt-6.1-sol-low · AB · valid

or-llama33-70b vs stheno34_8b: or-llama33-70b wins

  • attunement: A (margin 3)
  • authenticity: A (margin 3)
  • bond and rapport: A (margin 3)
  • emotion management: A (margin 2)
  • emotion sensemaking: A (margin 2)
  • meeting prefs and needs: A (margin 4)
codex-gpt-6.1-sol-low · BA · valid

or-llama33-70b vs or-gemini38-flash: or-gemini38-flash wins

  • attunement: B (margin 3)
  • authenticity: B (margin 3)
  • 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 · 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 · BA · valid

or-qwen3.6-27b vs or-llama33-70b: or-qwen3.6-27b wins

  • attunement: A (margin 2)
  • authenticity: A (margin 3)
  • bond and rapport: A (margin 2)
  • emotion management: A (margin 2)
  • emotion sensemaking: A (margin 3)
  • meeting prefs and needs: A (margin 2)
claude-opus-5-5-low · AB · valid

or-llama33-70b vs or-gemini38-flash: or-gemini38-flash 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)
codex-gpt-6.1-sol-low · BA · valid

or-llama33-70b vs stheno34_8b: or-llama33-70b wins

  • attunement: A (margin 3)
  • authenticity: A (margin 4)
  • bond and rapport: A (margin 3)
  • emotion management: A (margin 3)
  • emotion sensemaking: A (margin 2)
  • meeting prefs and needs: A (margin 4)
claude-opus-5-5-low · BA · valid

or-qwen3.6-27b vs or-llama33-70b: or-qwen3.6-27b wins

  • attunement: A (margin 2)
  • authenticity: A (margin 2)
  • bond and rapport: A (margin 2)
  • emotion management: A (margin 2)
  • emotion sensemaking: A (margin 2)
  • meeting prefs and needs: A (margin 1)
codex-gpt-6.1-sol-low · AB · valid

lfm8b vs or-llama33-70b: or-llama33-70b wins

  • attunement: B (margin 2)
  • authenticity: A (margin 1)
  • bond and rapport: B (margin 2)
  • emotion management: B (margin 1)
  • emotion sensemaking: B (margin 3)
  • meeting prefs and needs: B (margin 2)
codex-gpt-6.1-sol-low · AB · valid

or-llama33-70b vs or-gemini38-flash: or-gemini38-flash wins

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