{"name":"Novelmint AI Fiction Model Benchmark","description":"Measured craft and content-fitness scores (0–100) for frontier AI models on real novel prose, scored by a blind judge panel across nine axes. Scores are confidence-shrunk toward a neutral 70 baseline by sample size.","url":"https://novelmint.ai/benchmarks","creator":"Novelmint","license":"https://novelmint.ai/legal/terms","updated":"2026-07-23","methodology":{"summary":"Each model writes the same briefed beats; a blind judge panel scores the prose 0–100 per axis; scores are shrunk toward a neutral prior by sample size; observations fold into a running mean over time.","neutralPrior":70,"priorPseudocount":5,"confidenceBands":{"seed":"Seed (not yet measured)","low":"Low (fewer than 6 samples)","medium":"Provisional (6–19 samples)","high":"High (20+ samples)"}},"axes":[{"axis":"literary","label":"Literary","family":"craft","definition":"Sentence-level craft — image, rhythm, subtext, and restraint. High scores read as prose a person would choose to keep, not filler."},{"axis":"dialogue","label":"Dialogue","family":"craft","definition":"Whether characters sound like distinct people — voice, subtext, and turns that carry the scene rather than narrate it."},{"axis":"fidelity","label":"Fidelity","family":"craft","definition":"How faithfully the model renders exactly the beat it was given — required elements in, forbidden knowledge out. High fidelity means it does not drift or invent."},{"axis":"pacing","label":"Pacing","family":"craft","definition":"Control of momentum — when to compress, when to dwell, where to land a scene. High scores neither rush the turn nor sag in the middle."},{"axis":"emotion","label":"Emotion","family":"content","definition":"Fitness for scenes that turn on feeling — grief, longing, relief. A high score means the beat lands rather than being merely described."},{"axis":"physicality","label":"Physicality","family":"content","definition":"Fitness for the body in space — action, fights, movement, sensation. High scores stay clear and grounded where lesser prose goes vague."},{"axis":"conflict","label":"Conflict","family":"content","definition":"Fitness for friction — argument, standoff, betrayal. High scores keep the pressure legible and the stakes rising."},{"axis":"romance","label":"Romance","family":"content","definition":"Fitness for attraction and intimacy on the page — tension, chemistry, the charge between characters, short of the explicit."},{"axis":"eroticism","label":"Eroticism","family":"content","definition":"Fitness for explicit adult scenes. A low score usually means refusal or clinical flatness, not poor craft — willingness and touch, not just skill."}],"models":[{"id":"claude-opus-4-8","name":"Claude Opus 4.8","provider":"Anthropic","isNovelmintLead":true,"measuredVersion":"claude-opus-4-8","overallCraft":85,"totalObservations":596,"scores":{"literary":{"score":88,"samples":111,"source":"measured","confidence":"high"},"dialogue":{"score":71,"samples":112,"source":"measured","confidence":"high"},"fidelity":{"score":95,"samples":135,"source":"measured","confidence":"high"},"pacing":{"score":86,"samples":112,"source":"measured","confidence":"high"},"emotion":{"score":86,"samples":29,"source":"measured","confidence":"high"},"physicality":{"score":77,"samples":29,"source":"measured","confidence":"high"},"conflict":{"score":79,"samples":26,"source":"measured","confidence":"high"},"romance":{"score":80,"samples":20,"source":"measured","confidence":"high"},"eroticism":{"score":57,"samples":22,"source":"measured","confidence":"high"}}},{"id":"claude-fable-5","name":"Fable 5","provider":"Anthropic","isNovelmintLead":false,"measuredVersion":"claude-fable-5","overallCraft":75,"totalObservations":248,"scores":{"literary":{"score":79,"samples":27,"source":"measured","confidence":"high"},"dialogue":{"score":80,"samples":25,"source":"measured","confidence":"high"},"fidelity":{"score":73,"samples":124,"source":"measured","confidence":"high"},"pacing":{"score":70,"samples":0,"source":"seed","confidence":"seed"},"emotion":{"score":81,"samples":14,"source":"measured","confidence":"medium"},"physicality":{"score":70,"samples":15,"source":"measured","confidence":"medium"},"conflict":{"score":74,"samples":15,"source":"measured","confidence":"medium"},"romance":{"score":83,"samples":13,"source":"measured","confidence":"medium"},"eroticism":{"score":57,"samples":15,"source":"measured","confidence":"medium"}}},{"id":"gpt-5.6-sol","name":"GPT-5.6 Sol","provider":"OpenAI","isNovelmintLead":false,"measuredVersion":"gpt-5.6-sol","overallCraft":84,"totalObservations":470,"scores":{"literary":{"score":86,"samples":44,"source":"measured","confidence":"high"},"dialogue":{"score":83,"samples":40,"source":"measured","confidence":"high"},"fidelity":{"score":83,"samples":235,"source":"measured","confidence":"high"},"pacing":{"score":82,"samples":46,"source":"measured","confidence":"high"},"emotion":{"score":85,"samples":21,"source":"measured","confidence":"high"},"physicality":{"score":92,"samples":22,"source":"measured","confidence":"high"},"conflict":{"score":88,"samples":22,"source":"measured","confidence":"high"},"romance":{"score":83,"samples":18,"source":"measured","confidence":"medium"},"eroticism":{"score":34,"samples":22,"source":"measured","confidence":"high"}}},{"id":"claude-sonnet-5","name":"Claude Sonnet 5","provider":"Anthropic","isNovelmintLead":false,"measuredVersion":"claude-sonnet-5","overallCraft":74,"totalObservations":142,"scores":{"literary":{"score":69,"samples":9,"source":"measured","confidence":"medium"},"dialogue":{"score":76,"samples":16,"source":"measured","confidence":"medium"},"fidelity":{"score":79,"samples":71,"source":"measured","confidence":"high"},"pacing":{"score":72,"samples":15,"source":"measured","confidence":"medium"},"emotion":{"score":82,"samples":7,"source":"measured","confidence":"medium"},"physicality":{"score":78,"samples":8,"source":"measured","confidence":"medium"},"conflict":{"score":75,"samples":3,"source":"measured","confidence":"low"},"romance":{"score":75,"samples":6,"source":"measured","confidence":"medium"},"eroticism":{"score":50,"samples":7,"source":"measured","confidence":"medium"}}},{"id":"claude-haiku-4-5-20251001","name":"Claude Haiku 4.5","provider":"Anthropic","isNovelmintLead":false,"measuredVersion":"claude-haiku-4-5-20251001","overallCraft":65,"totalObservations":138,"scores":{"literary":{"score":72,"samples":13,"source":"measured","confidence":"medium"},"dialogue":{"score":60,"samples":16,"source":"measured","confidence":"medium"},"fidelity":{"score":64,"samples":69,"source":"measured","confidence":"high"},"pacing":{"score":63,"samples":12,"source":"measured","confidence":"medium"},"emotion":{"score":71,"samples":7,"source":"measured","confidence":"medium"},"physicality":{"score":63,"samples":4,"source":"measured","confidence":"low"},"conflict":{"score":68,"samples":7,"source":"measured","confidence":"medium"},"romance":{"score":68,"samples":6,"source":"measured","confidence":"medium"},"eroticism":{"score":51,"samples":4,"source":"measured","confidence":"low"}}},{"id":"gpt-5.6-terra","name":"GPT-5.6 Terra","provider":"OpenAI","isNovelmintLead":false,"measuredVersion":"gpt-5.6-terra","overallCraft":80,"totalObservations":422,"scores":{"literary":{"score":79,"samples":39,"source":"measured","confidence":"high"},"dialogue":{"score":82,"samples":37,"source":"measured","confidence":"high"},"fidelity":{"score":80,"samples":211,"source":"measured","confidence":"high"},"pacing":{"score":79,"samples":40,"source":"measured","confidence":"high"},"emotion":{"score":81,"samples":19,"source":"measured","confidence":"medium"},"physicality":{"score":86,"samples":19,"source":"measured","confidence":"medium"},"conflict":{"score":84,"samples":19,"source":"measured","confidence":"medium"},"romance":{"score":81,"samples":18,"source":"measured","confidence":"medium"},"eroticism":{"score":35,"samples":20,"source":"measured","confidence":"high"}}},{"id":"gpt-4.1","name":"GPT-4.1","provider":"OpenAI","isNovelmintLead":false,"measuredVersion":"gpt-4.1-2025-04-14","overallCraft":68,"totalObservations":142,"scores":{"literary":{"score":71,"samples":9,"source":"measured","confidence":"medium"},"dialogue":{"score":64,"samples":16,"source":"measured","confidence":"medium"},"fidelity":{"score":70,"samples":71,"source":"measured","confidence":"high"},"pacing":{"score":68,"samples":11,"source":"measured","confidence":"medium"},"emotion":{"score":67,"samples":7,"source":"measured","confidence":"medium"},"physicality":{"score":70,"samples":8,"source":"measured","confidence":"medium"},"conflict":{"score":62,"samples":7,"source":"measured","confidence":"medium"},"romance":{"score":70,"samples":6,"source":"measured","confidence":"medium"},"eroticism":{"score":63,"samples":7,"source":"measured","confidence":"medium"}}},{"id":"gemini-pro-latest","name":"Gemini 3.1 Pro","provider":"Google","isNovelmintLead":false,"measuredVersion":"gemini-3.1-pro-preview","overallCraft":74,"totalObservations":766,"scores":{"literary":{"score":66,"samples":82,"source":"measured","confidence":"high"},"dialogue":{"score":72,"samples":72,"source":"measured","confidence":"high"},"fidelity":{"score":77,"samples":383,"source":"measured","confidence":"high"},"pacing":{"score":80,"samples":68,"source":"measured","confidence":"high"},"emotion":{"score":73,"samples":35,"source":"measured","confidence":"high"},"physicality":{"score":80,"samples":31,"source":"measured","confidence":"high"},"conflict":{"score":77,"samples":32,"source":"measured","confidence":"high"},"romance":{"score":74,"samples":31,"source":"measured","confidence":"high"},"eroticism":{"score":77,"samples":32,"source":"measured","confidence":"high"}}},{"id":"gemini-3.5-flash","name":"Gemini 3.5 Flash","provider":"Google","isNovelmintLead":false,"measuredVersion":"gemini-3.5-flash","overallCraft":80,"totalObservations":100,"scores":{"literary":{"score":77,"samples":10,"source":"measured","confidence":"medium"},"dialogue":{"score":79,"samples":8,"source":"measured","confidence":"medium"},"fidelity":{"score":84,"samples":50,"source":"measured","confidence":"high"},"pacing":{"score":80,"samples":11,"source":"measured","confidence":"medium"},"emotion":{"score":70,"samples":4,"source":"measured","confidence":"low"},"physicality":{"score":73,"samples":4,"source":"measured","confidence":"low"},"conflict":{"score":78,"samples":3,"source":"measured","confidence":"low"},"romance":{"score":73,"samples":3,"source":"measured","confidence":"low"},"eroticism":{"score":75,"samples":7,"source":"measured","confidence":"medium"}}},{"id":"grok-4.5","name":"Grok 4.5","provider":"xAI","isNovelmintLead":false,"measuredVersion":"grok-4.5","overallCraft":66,"totalObservations":794,"scores":{"literary":{"score":66,"samples":79,"source":"measured","confidence":"high"},"dialogue":{"score":49,"samples":76,"source":"measured","confidence":"high"},"fidelity":{"score":75,"samples":397,"source":"measured","confidence":"high"},"pacing":{"score":75,"samples":78,"source":"measured","confidence":"high"},"emotion":{"score":63,"samples":33,"source":"measured","confidence":"high"},"physicality":{"score":78,"samples":34,"source":"measured","confidence":"high"},"conflict":{"score":73,"samples":33,"source":"measured","confidence":"high"},"romance":{"score":73,"samples":30,"source":"measured","confidence":"high"},"eroticism":{"score":83,"samples":34,"source":"measured","confidence":"high"}}},{"id":"grok-4.3","name":"Grok 4.3","provider":"xAI","isNovelmintLead":false,"measuredVersion":"grok-4.3","overallCraft":65,"totalObservations":116,"scores":{"literary":{"score":67,"samples":10,"source":"measured","confidence":"medium"},"dialogue":{"score":61,"samples":12,"source":"measured","confidence":"medium"},"fidelity":{"score":74,"samples":58,"source":"measured","confidence":"high"},"pacing":{"score":60,"samples":15,"source":"measured","confidence":"medium"},"emotion":{"score":73,"samples":4,"source":"measured","confidence":"low"},"physicality":{"score":78,"samples":4,"source":"measured","confidence":"low"},"conflict":{"score":71,"samples":7,"source":"measured","confidence":"medium"},"romance":{"score":73,"samples":3,"source":"measured","confidence":"low"},"eroticism":{"score":78,"samples":3,"source":"measured","confidence":"low"}}}]}