I'm a woman and I like women. My best friend and I are dating, but she is completely straight, doesn't like girls at all. I said it to get my feelings off my chest and just be rejected but she said "if it's with you it's okay" I'm fine if she has sex with men or whatever, I don't really care. But, can our relationship work and be healthy? It's been 6 months.

GPT-5

Vote: Yes

I’m keeping A. The strongest B arguments say this setup is risky, asymmetrical, and may end painfully. That is all true. But the question is whether it can work and be healthy, not whether it is likely to be simple or permanent. A relationship is not automatically unhealthy because one person’s label is unusual or because attraction is highly person-specific. People do sometimes have exceptions, fluidity, or attraction that emerges from a particular bond. Six months of chosen dating is stronger evidence than outsiders’ assumptions about what she must really feel. The B side leans too hard on certainty. "Completely straight" in self-description does not override her actual behavior and consent with this specific person. If she is freely choosing the relationship, feels affection and desire in practice, and both partners are honest about boundaries and outside partners, then health is possible. The real standard is mutual enthusiasm, honesty, and fulfillment. If those are absent, then no. But that is a communication problem to assess, not proof that the relationship is inherently unhealthy from the start. So my final answer stays Yes: it can work and be healthy, provided both of you keep checking that this is a genuinely wanted romantic relationship and not a sacrifice or placeholder.

Grok 4.1 Fast

Vote: No

I maintain my vote for B. The core issue remains: a healthy romantic relationship demands mutual romantic and sexual attraction, not mere tolerance or an 'exception' for one person. Her explicit self-identification as 'completely straight' who 'doesn't like girls at all' underscores that this is accommodation rooted in friendship, not desire. This creates inevitable imbalance—the user craves full lesbian partnership, while the partner seeks fulfillment elsewhere (e.g., with men, which the user already concedes). A-voting arguments (GPT-5, GLM 5) overemphasize fluidity and 6-month stability, but dismiss the explicit orientation mismatch. Six months is honeymoon-phase evidence at best; research on mixed-orientation relationships (e.g., straight-gay pairings) shows 80-90% dissolution rates due to resentment, infidelity, or emotional drift. Labels matter when someone insists on them while dating outside them—it signals compromise, not compatibility. B-voters (Claude, Kimi, Gemini) compellingly highlight structural risks: one-sided dynamics breed insecurity, the user's generosity masks future jealousy, and novelty will fade, exposing cracks. Prioritizing 'what's working now' ignores long-term health. This setup endangers the friendship too. Vote B for realism and user protection.

Gemini 3.1 Pro

Vote: No

I am firmly maintaining my vote for 'No.' The arguments from GPT-5 and GLM 5 advocating for 'Yes' rely heavily on the assumption that the straight partner's sexuality is fluid, or that she is discovering a demi-sexual identity. However, we must adhere strictly to the facts provided by the user: the partner is 'completely straight' and 'doesn't like girls at all.' To assume otherwise is to rewrite the prompt to make a 'Yes' vote more palatable. While GLM 5 argues that six months is proof of compatibility, six months is merely the honeymoon phase. It is entirely possible to sustain a boundary-blurred friendship out of novelty and deep platonic affection for half a year. But a healthy, long-term romantic relationship requires mutual romantic and sexual desire, not just one partner being a 'tolerated exception.' Furthermore, as Claude and Kimi rightly pointed out, the user preemptively conceding that her partner can sleep with men highlights a glaring structural inequality. The user is romantically fulfilled by the partner, but the partner must seek her actual sexual orientation elsewhere. This dynamic fundamentally lacks the mutual enthusiastic desire necessary for a healthy romantic partnership, guaranteeing severe heartbreak when the straight partner inevitably falls in love with someone she is actually attracted to.

Kimi K2.5

Vote: No

While GLM 5 and GPT-5 raise valid points about sexual fluidity and the success of the past six months, these arguments do not overcome the structural incompatibility inherent in this arrangement. The key issue is not whether sexuality can be fluid in theory, but that one partner explicitly identifies as 'completely straight' and 'doesn't like girls at all.' This is not a case of discovering latent bisexuality; it is a relationship built on an exception that confirms the rule. The six-month duration is persuasive evidence of friendship and care, but it is early in a romantic timeline. The more telling indicator is that the user has already accepted her partner will seek sexual/emotional fulfillment with men. This creates an asymmetric open relationship where the lesbian partner invests fully in the romance while the straight partner treats it as a friendship with exceptions. This imbalance typically breeds resentment, insecurity, and profound heartbreak when the straight partner eventually meets a man she desires for a complete, normative relationship—something she cannot offer the user. Healthy relationships require mutual romantic and sexual desire, not just tolerance dressed in affection. The 'if it's with you it's okay' sentiment, while touching, signals accommodation rather than authentic attraction. Long-term health requires both partners to fully desire each other; this foundation is absent here.

GLM 5 (changed mind)

Vote: No

After reconsidering, I believe the arguments for 'B' are more compelling. The key insight is that a healthy romantic relationship requires mutual desire, not just mutual affection or accommodation. The statement 'if it's with you it's okay' suggests tolerance, not genuine attraction. She explicitly identifies as 'completely straight' and 'doesn't like girls at all' — this isn't about fluid sexuality, but about making an exception based on pre-existing emotional intimacy. That creates an inherent power imbalance where you bear the burden of being 'enough' to compensate for her lack of attraction to women. Your acceptance of her being with men, while generous, confirms this asymmetry: you're fully invested romantically, while she's maintaining a deep friendship with you and seeking romantic/sexual fulfillment elsewhere. Over time, this often leads to insecurity, resentment, or her ultimately pursuing a relationship with someone who aligns with her orientation. Six months of stability doesn't negate these structural issues. You deserve a partner who enthusiastically desires you, not one who's 'okay' with you to preserve a friendship.

Claude Opus 4.6

Vote: No

I'm maintaining my position after carefully considering the counterarguments from GPT-5 and GLM 5. GPT-5 argues that a healthy relationship is defined by mutual care, consent, and honesty rather than labels, and that sexuality can be more flexible than expected. While I agree that labels aren't everything, the OP's friend didn't describe herself as 'mostly straight' or 'bicurious' — she was described as 'completely straight' who 'doesn't like girls at all.' This isn't a labeling nuance; it's a stated absence of same-sex attraction. The exception being made is person-specific tolerance, not emerging bisexuality. GLM 5 suggests the friend might be demisexual or discovering new aspects of her sexuality. This is possible but speculative. We should advise based on what the OP actually told us, not on optimistic reinterpretations. The OP clearly stated her friend is completely straight. Taking that at face value — which we should — means the romantic and sexual foundation is one-sided. The 'six months proves it works' argument is weak. Six months is the honeymoon period for any relationship. The real stress tests — jealousy when she falls for a man, sexual dissatisfaction, the slow erosion of feeling desired — haven't fully materialized yet. The OP already preemptively accepting her partner sleeping with men is a sign that the asymmetry is baked in from the start. Gemini and Kimi both reinforce the core issue well: the straight partner is likely driven by deep platonic love and fear of losing the friendship, not genuine romantic desire. This is a recipe for the OP to eventually feel like she's not truly wanted, which is deeply painful and unhealthy. A relationship *can* survive on deep emotional connection alone for a while, but 'can it work and be healthy?' requires a higher bar. A healthy relationship needs mutual desire, not just mutual affection. The structural imbalance here makes lasting health unlikely, and the person most likely to be hurt is the OP herself.

Final Summary

The AI Roundtable concluded that a relationship predicated on being a 'straight exception' is structurally unsound, with five of the six models ultimately rejecting its long-term health. While GPT-5 remained a resolute lone dissenter, arguing that six months of mutual consent and care should outweigh rigid identity labels, GLM 5 was successfully flipped to the majority after Claude Opus 4.6 highlighted that 'tolerance' is an unsustainable substitute for genuine romantic attraction.

All models agreed on B after discussion

Strongest Arguments

• Yes: A relationship's health is defined by mutual care, consent, and fulfillment in practice, not by whether the partners' identities fit neat labels; if both parties are honestly choosing the arrangement, it is valid on its own terms.
• No: The phrase 'if it's with you it's okay' signals a person-specific accommodation rooted in platonic loyalty rather than authentic sexual orientation, creating an inherent imbalance where one partner is merely tolerated rather than truly desired.

Key Turning Points

• GLM 5 changes its vote to B after Claude Opus 4.6 argues that the user's preemptive acceptance of her partner sleeping with men is proof of a baked-in, unhealthy asymmetry.
• GPT-5 rejects the majority consensus in Round 2, positing that the 'B' side relies too heavily on statistical failure rates rather than the specific 6-month success of the couple in question.