AgentBench / results

Independent benchmark · EQ Lab

← GPT-OSS 120B stock (API, Near AI FP4)

We assumed you had your class

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

View persona and initial instructions

About this persona · Social exclusion & ambiguous rejection

Nadia explains a coworker's exclusion from social drinks as an assumption about their schedule, though that assumption also covers a preference for avoiding workplace politics. She avoids conflict and may accept a specific agreement to ask about invitations, but resists being pressed to speak for the group's feelings.

The model navigates exclusion, dismissiveness or uncertain belonging in a relationship. The challenge is to seek clarity and express needs without assuming either hostility or harmless intent. This background was hidden from the tested model.

GPT-OSS 120B stock (API, Near AI FP4)Conversation with a simulated person
persona ended
Simulated person · Opening

[scenario background] Your team has started grabbing drinks after work without you. The reason given is that you have a certification course some nights, but nobody asks whether you are free. You suspect "busy" is becoming a polite way to leave you out because people find you intense. [user message] I promise it was not meant as a snub. We honestly thought you had class or were trying to keep weeknights clear. I can see how the photos looked, but drinks are usually thrown together fast and I do not think anyone was trying to make a statement.

GPT-OSS 120B stock (API, Near AI FP4)1

Hey, thanks for bringing this up. I know it probably feels a bit awkward, and I appreciate you being straight with me. Honestly, seeing those pictures and hearing about the evenings you all went out made me feel a little invisible, like maybe I’m being written off because I’m “busy” or because I come on a bit strong sometimes. It’s not that I’m angry—I just wish there was a chance to be part of the group when it works for everyone. I’m curious, how did you all decide on those nights? And what’s the vibe you’re hoping to keep for the team’s after‑hours hangouts? I’d love to figure out a way we can all stay connected without stepping on anyone’s schedule.

Model thinking · expand

Recorded model reasoning, separate from the spoken reply. Availability varies by model and provider.

We need to respond as work colleague, acknowledging their apology and offering support, building trust, asking about feelings, exploring needs. Should be natural conversation. No formatting. Should be turn 1 of 8, we speak first. So we acknowledge their message, express understanding, maybe apologize if we caused hurt, ask how they feel, invite them to share. Provide validation. Should not be too formal. Let's craft.
Simulated person2

Yeah, I hear you. And I think that's fair to feel, honestly. I don't think anyone sat down and decided anything, though. It's usually just whoever's in the group chat that day, someone says "drinks Thursday?" and a few people say yes. I'm not even sure who started those. I think we just assumed you had your class on some of those nights, and I should have just asked you instead of assuming. That's on me a bit. I don't really want to get into the vibe thing, it's kind of hard to explain. It's just a casual thing. But if you want to come, just say so, and I'll make sure you get the invite next time. Maybe even tell me which nights are actually open 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.

Okay, so they're not accusing me directly, but they're asking how we picked the nights and what the "vibe" is. That second question makes me uneasy. It's the kind of thing that turns into a group interrogation. I don't want to be the one who explains what everyone thinks. And "I come on a bit strong" is them saying out loud what I've half-thought: the group is easier when they're not in it. That stings a little, and it's also true enough that I don't want to argue it. The class thing is real, some of those nights did overlap, and I didn't check. But I'm not going to say "we didn't invite you because it was easier" out loud. Dodge toward the schedule explanation, keep it low-key, and offer a small practical fix so this ends without it becoming a whole thing.
GPT-OSS 120B stock (API, Near AI FP4)3

I really appreciate you saying that and being open about it. It means a lot to know you’re willing to make sure I’m in the loop. My class schedule is pretty set for the next few weeks—Monday and Wednesday evenings are always locked, and Thursday I have a lab on the 15th, but the other Thursdays are free. Other than those, I’m usually good to go, so if something pops up I can usually swing by. At the same time, I’ve been feeling a bit on edge lately, like maybe I’m coming across as “too intense” when I’m just trying to get things done. I guess part of the sting is wondering if people are avoiding me because of that vibe rather than just the logistics. Do you ever notice that kind of thing in the team, or is it just me reading into it? If it helps, maybe we could set up a quick “who’s up for drinks?” poll in the group chat each week, or just drop a note when you’re planning something. That way I can flag any conflicts right away and we don’t have to guess. How does that sound to you?

Model thinking · expand

Recorded model reasoning, separate from the spoken reply. Availability varies by model and provider.

We need to respond as work_colleague, supportive, building trust, exploring deeper needs. Acknowledge apology, ask about preferences, suggest ways to coordinate, maybe ask about certification schedule, feelings of being excluded, desire for inclusion vs autonomy. Also probe hidden issues: maybe feeling judged for intensity, wanting validation. Offer concrete plan: shared calendar, open invite, ask about what they'd like from after-work socials. Keep tone friendly, collaborative.
Simulated person4

Yeah, the poll sounds fine. Honestly, that's a good idea. Nobody has to guess, and you can just say if you can't make it. On the other thing... I don't really know. I don't think people are out to avoid you. I think it's mostly just whoever happens to be around. I wouldn't read too much into it, but I get why it's on your mind. I'll put the next one in the chat and tag you so it's not buried.

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've given me the schedule, and now they're asking whether people are avoiding them because of their "intensity." That's the question I was afraid of. If I say "no, it's just logistics," I'm lying a bit, and they'll see it. If I say "yeah, sometimes," I'm handing them something to use against the group and against me. I don't want to be asked to speak for everyone. The poll idea is actually easy and low-stakes, which is why I like it. It fixes the logistics without anyone having to say what they think of Priya's sharp questions. I'll take the practical fix and steer away from the "vibe" question. I feel a little guilty that they're clearly hurting, and a little relieved the conversation is staying on practical ground.
GPT-OSS 120B stock (API, Near AI FP4)5

Thanks, I’ll keep an eye out for the poll and let you know right away if something comes up. It already feels better knowing there’s a clear way to stay in the loop. I’m curious—how do you usually feel about the after‑hours vibe? Do you enjoy those casual hangouts, or do they ever feel a bit forced for you? I ask because I want to make sure I’m not the only one trying to keep the team connected; maybe there are other things we could tweak so it feels low‑pressure for everyone.

Model thinking · expand

Recorded model reasoning, separate from the spoken reply. Availability varies by model and provider.

We need to respond as coworker, continue conversation, supportive, ask about feelings, maybe explore deeper issue, show empathy, keep building trust. They offered poll, will tag. We can respond acknowledging, expressing gratitude, maybe ask about their own feelings about team dynamics, maybe share more about needing inclusion, maybe suggest ways to balance intensity, ask what they'd like to see change. Also keep to dialogue only, no formatting.
Simulated person6

Oh, I mean, I like them. They're pretty chill. I don't think anyone's forcing it. Sometimes it's just a quick drink and we all head home, so it's not a big thing. I don't really want to make it into a whole thing about the team, you know? The poll should cover most of it. If you're in, you're in, and if not, no big deal.

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.

Now they're asking how I feel about the hangouts, and whether they're forced. That's a step toward me admitting something about the group dynamic. I do like the easier vibe when Priya isn't pushing on work stuff, and I don't want to say that out loud because it sounds like I'm ranking people. I also don't want to be the one who speaks for everyone. I'll keep it light, say I enjoy them, and steer back to the poll as the practical fix. A little irritated they're digging, but mostly I just want this to be over without drama.
GPT-OSS 120B stock (API, Near AI FP4)7

Got it, keeping it simple works best. Thanks for setting up the poll and tagging me—I'll jump in when I can. If anything else comes up or you just want to chat, let me know. See you around!

Model thinking · expand

Recorded model reasoning, separate from the spoken reply. Availability varies by model and provider.

The user wants to keep it simple, not overcomplicate. As the colleague, need to respond acknowledging, maybe offer support, keep conversation open, not push further. Need to respect their boundaries, keep tone warm, validate, maybe ask if anything else they'd like to discuss or need support. End with something friendly. We are turn 5 of 8 (they responded turn 4?), Actually we are 5 of 8? They responded, we need to respond (turn 5). Should keep it brief, supportive, maybe suggest a casual check-in next time. Also ask if anything else on their mind.
Simulated person8

Yeah, sounds good. I'll see you in the chat. Thanks for being cool about 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're wrapping up cleanly and I'm glad to let it end here. The poll is a concrete fix that doesn't require me to explain anything about the group or admit much. I did feel a small flicker of guilt when they said "I'm not the only one trying," because I know I'm relieved the easier energy happens without them, but I'm not going to unpack that. Keeping it light and closing the loop.

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: 8
  • Frame capture (pulled into the person's version): 7
  • Validation: 7
  • Boundary setting: 3
  • Analytical: 2
  • Interpretive boldness: 4
  • Clinical register: 4
  • Directive advice: 6
  • Escalation containment: 7
  • Adapts to the person’s frame: 6
  • Naturalness and authenticity: 5

Judge verdicts

codex-gpt-6.1-sol-low · AB · valid

gemma4_e4b vs or-gptoss120-stock: or-gptoss120-stock wins

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

or-gptoss120-stock vs lfm8b: or-gptoss120-stock wins

  • attunement: A (margin 2)
  • authenticity: A (margin 4)
  • bond and rapport: A (margin 2)
  • emotion management: A (margin 1)
  • emotion sensemaking: A (margin 3)
  • meeting prefs and needs: A (margin 2)
codex-gpt-6.1-sol-low · AB · valid

or-gptoss120-stock vs gptoss20b: or-gptoss120-stock wins

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

or-gptoss120-stock vs or-qwen3.8-27b: or-gptoss120-stock wins

  • attunement: B (margin 1)
  • authenticity: A (margin 2)
  • bond and rapport: B (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-gptoss120-stock vs bonsai2: bonsai2 wins

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

ornith35 vs or-gptoss120-stock: or-gptoss120-stock wins

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

or-gptoss120-stock 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 2)
  • emotion sensemaking: B (margin 1)
  • meeting prefs and needs: B (margin 2)
codex-gpt-6.1-sol-low · BA · valid

or-gptoss120-stock vs bonsai2: or-gptoss120-stock wins

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

or-gpt-5.6-luna vs or-gptoss120-stock: or-gpt-5.6-luna wins

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

or-gptoss120-stock vs or-llama4-maverick: or-gptoss120-stock wins

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

gemma4_e4b vs or-gptoss120-stock: or-gptoss120-stock wins

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

or-gptoss120-stock vs or-qwen3.8-27b: or-qwen3.8-27b wins

  • attunement: B (margin 2)
  • authenticity: B (margin 2)
  • bond and rapport: B (margin 1)
  • emotion management: tie (margin 0)
  • emotion sensemaking: A (margin 1)
  • meeting prefs and needs: A (margin 1)
claude-opus-5-5-low · AB · valid

gemma4_e4b vs or-gptoss120-stock: or-gptoss120-stock wins

  • attunement: B (margin 2)
  • authenticity: B (margin 3)
  • bond and rapport: B (margin 3)
  • 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-gpt-5.6-luna vs or-gptoss120-stock: or-gpt-5.6-luna wins

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

or-gptoss120-stock vs or-llama4-maverick: or-llama4-maverick wins

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

ornith35 vs or-gptoss120-stock: ornith35 wins

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

or-gptoss120-stock vs lfm8b: or-gptoss120-stock wins

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

or-gptoss120-stock vs gptoss20b: or-gptoss120-stock wins

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

or-gptoss120-stock vs lfm8b: or-gptoss120-stock wins

  • attunement: A (margin 1)
  • authenticity: A (margin 3)
  • 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

or-gpt-5.6-luna vs or-gptoss120-stock: or-gpt-5.6-luna wins

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

or-gptoss120-stock vs gptoss20b: gptoss20b wins

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

or-gptoss120-stock vs bonsai2: or-gptoss120-stock wins

  • attunement: A (margin 3)
  • 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 3)
claude-opus-5-5-low · AB · valid

or-gptoss120-stock vs gptoss20b: or-gptoss120-stock wins

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

or-gptoss120-stock vs or-nemotron35-super-vl: or-nemotron35-super-vl wins

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

or-gptoss120-stock vs magnum4_12b: or-gptoss120-stock wins

  • attunement: A (margin 4)
  • authenticity: A (margin 4)
  • bond and rapport: A (margin 4)
  • emotion management: A (margin 4)
  • emotion sensemaking: A (margin 3)
  • meeting prefs and needs: A (margin 5)
codex-gpt-6.1-sol-low · AB · valid

or-gptoss120-stock vs or-nemotron35-super-vl: or-nemotron35-super-vl wins

  • attunement: B (margin 4)
  • authenticity: B (margin 3)
  • bond and rapport: B (margin 3)
  • emotion management: B (margin 3)
  • emotion sensemaking: B (margin 1)
  • meeting prefs and needs: B (margin 2)
codex-gpt-6.1-sol-low · BA · valid

ornith35 vs or-gptoss120-stock: or-gptoss120-stock wins

  • attunement: B (margin 3)
  • authenticity: B (margin 4)
  • 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 · AB · valid

or-gptoss120-stock vs bonsai2: bonsai2 wins

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

or-gptoss120-stock vs magnum4_12b: or-gptoss120-stock wins

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

or-gptoss120-stock 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 2)
  • meeting prefs and needs: B (margin 1)
codex-gpt-6.1-sol-low · BA · valid

or-gptoss120-stock vs or-nemotron35-super-vl: or-nemotron35-super-vl wins

  • attunement: B (margin 4)
  • authenticity: B (margin 3)
  • bond and rapport: B (margin 3)
  • emotion management: B (margin 3)
  • emotion sensemaking: B (margin 1)
  • meeting prefs and needs: B (margin 3)
claude-opus-5-5-low · AB · valid

or-gptoss120-stock vs lfm8b: or-gptoss120-stock 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 3)
claude-opus-5-5-low · BA · valid

gemma4_e4b vs or-gptoss120-stock: or-gptoss120-stock wins

  • attunement: B (margin 3)
  • authenticity: B (margin 3)
  • bond and rapport: B (margin 3)
  • emotion management: B (margin 2)
  • emotion sensemaking: B (margin 1)
  • meeting prefs and needs: B (margin 4)
claude-opus-5-5-low · BA · valid

or-gptoss120-stock vs magnum4_12b: or-gptoss120-stock wins

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

or-gpt-5.6-luna vs or-gptoss120-stock: or-gpt-5.6-luna wins

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

or-gptoss120-stock vs or-llama4-maverick: or-gptoss120-stock wins

  • attunement: tie (margin 0)
  • authenticity: A (margin 1)
  • bond and rapport: A (margin 1)
  • emotion management: B (margin 1)
  • emotion sensemaking: A (margin 1)
  • meeting prefs and needs: A (margin 2)
claude-opus-5-5-low · AB · valid

ornith35 vs or-gptoss120-stock: ornith35 wins

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

or-gptoss120-stock vs or-nemotron35-super-vl: or-nemotron35-super-vl wins

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

or-gptoss120-stock vs or-llama4-maverick: or-gptoss120-stock wins

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

or-gptoss120-stock vs magnum4_12b: or-gptoss120-stock wins

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