THIS HAS BEEN UPDATED TO: https://rentry.org/LLM-Compare/
I used ChatGPT to generate this but still suffers some bias as it's made for my need for dark fantasy x dark humour. Featherless API. Rating 1-3
| Model Name | Model Type | Context Length | Roleplay Style | Best For | Bad At | Horniness Level | Positivity Bias | Pepper's Notes |
|---|---|---|---|---|---|---|---|---|
| Steelskull/L3.3-Nevoria-R1-70b | LLaMA | 16K | Highly immersive storytelling, deep narrative consistency | Dark fantasy, long-term RP, character-driven interactions | Can be too rigid for casual or comedic RP | 1 | 2 | My fave with a good balance, does comedy decently if example dialogue sets the tone |
| Steelskull/L3.3-San-Mai-R1-70b v0.5.OG | LLaMA | 16K | Deep worldbuilding, persistent memory, well-suited for dark settings | Serious plots, worldbuilding-heavy interactions | Requires commitment to in-depth worldbuilding | 2 | 2 | |
| Steelskull/L3.3-Cu-Mai-R1-70b | LLaMA | 16K | Balanced realism and fantasy elements, immersive storytelling | Dark fantasy, character-driven narratives | Can be too serious for lighthearted themes | 2 | 2 | |
| Steelskull/L3.3-Damascus-R1 | LLaMA | 16K | Dynamic, chaotic RP with strong thematic commitment | Chaotic, unpredictable RP, good for reactive storytelling | Not suited for slow-paced, structured RP | 3 | 1 | |
| Qwen/QwQ-32B | Qwen | 16K | Highly structured, maintains character cards well | Best for roleplay with established rules and restrictions | May struggle with freeform, spontaneous RP | 1 | 3 | |
| TheDrummer/Fallen-Llama-3.3-R1-70B-v1 | LLaMA | 16K | Dark and intense storytelling, immersive lore | Psychological depth, serious narratives | Not ideal for lighthearted or comedic RP | 3 | 1 | This one outright freaking hates user and I love it. Negativity bias. |
| EVA-UNIT-01/EVA-LLaMA-3.33-70B-v0.0 | LLaMA | 16K | Highly flexible, detailed responses, good for structured and freeform RP | Long-form RP, immersive storytelling | Not ideal for light, comedic RP | 1 | 3 | |
| AuriAetherwiing/MN-12B-Starcannon-v2 | Mistral | 16K | Highly creative, engaging narratives, suitable for long-term RP | Long-term character interactions, emotionally-driven RP | May produce overly verbose responses if not controlled | 2 | 2 | |
| EVA-UNIT-01/EVA-Qwen2.5-32B-v0.2 | Qwen 2.5 32b | 16K | Structured, adheres to character cards while allowing flexibility | Balanced RP with structured adherence to character settings | Not as good for very fast-paced interactions | 1 | 3 | It's been incoherent lately for some reason |
| LatitudeGames/Wayfarer-12B | Mistral Nemo 12b | 16K | Sandbox storytelling, open-ended interactions with creative input | Open-world, player-driven roleplay scenarios | Prefers non-linear storytelling over rigid scripts | 1 | 1 | 70B version is most stable for long term roleplay but a bit bland. |
| mistralai/Mistral-7B-Instruct-v0.2 | Mistral | 16K | Quick responses, good for casual RP but lacks deep memory | Quick, casual RP interactions | Struggles with maintaining deep memory over long sessions | 1 | 3 | |
| Nymbo/cohere-command-r | Cohere | 16K | Logical storytelling, useful for structured RP | Structured storytelling, professional-style RP | May feel too structured for freeform RP | 3 | 3 | |
| alpindale/WizardLM-2-8x22B | WizardLM | 16K | Fantasy and worldbuilding focused, deep lore handling | Detailed fantasy settings, lore-heavy worlds | Not suited for casual, quick RP | 1 | 3 | |
| TheDrummer/UnslopNemo-12B-v4.1 | Nemo | 16K | Immersive roleplay with a strong character consistency | Consistent character memory, RP that requires long-term engagement | Needs consistent engagement to work effectively | 3 | 2 | |
| TheDrummer/Rocinante-12B-v1.1 | Rocinante | 16K | Fast-paced interactions, good for action-oriented RP | Fast-paced adventure, episodic RP | Not suited for introspective, slow RP | 3 | 1 | Amazing for its size and for long term RP |
| inflatebot/MN-12B-Mag-Mell-R1 | Nemo | 16K | Deep memory and world continuity, strong adaptive capabilities | Worldbuilding with evolving continuity | Requires set-up to fully utilize worldbuilding features | 2 | 2 | |
| allura-org/Qwen2.5-32b-RP-Ink | Qwen | 16K | Creative but stable RP experiences, supports a variety of tones | Flexible creativity while maintaining structured RP flow | May sometimes lose deep contextual engagement | 2 | 3 | |
| Sao10k: llama 3.1 Euryale 70B V2.2 | LLaMA 3.3 | 16K | Psychological and immersive, good for deep character interactions | Psychological depth, slow-burn interactions | Requires depth to shine, not for surface-level RP | 1 | 1 | Love this for angst |
| Sao10K/14B-Qwen2.5-Kunou-v1 | Qwen | 16K | Well-balanced between humor and dark themes, can be highly adaptive | Dark humor and dark fantasy with immersive narratives | Not ideal for overly simplistic RP | 2 | 2 | |
| DeepSeek-AI/DeepSeek-R1 | DeepSeek | 32K | Highly structured, logical but adaptable storytelling | Highly structured roleplay, long-form worldbuilding | Not ideal for light, comedic RP | 1 | 2 | A little too wild for me, but great for short injections |
| trashpanda-org/QwQ-32B-Snowdrop-v0 | Qwen | 16K | Good prose and natural language, suited for natural and flowing RP | Flowing dialogue and immersive storytelling | Can feel too verbose if not guided properly | 1 | 3 | Great for prose-heavy interactions, feels natural and flowing |
| DatToad/Chuluun-Qwen2.5-72B-v0.08 | Qwen2.5 | 16K | Highly creative, character-intelligent with expressive prose | Emotionally complex storytelling, dark themes, unique tone shaping | May require temp tuning and manual editing for coherence | 3 | 2 | Surprisingly stable for a Qwen merge. Avoids common LLM-isms. One of my favourites for prose alongside Snowdrop. |
| Sao10K/Llama-3.3-70B-Vulpecula-r1 | LLaMA 3.3 | 16K | Creative, emotionally aware, flirty but grounded roleplay | Stable writing with romantic or intimate dynamics | May soften darker tones unless directed | 2 | 3 | Writing of Hanami 2.0. Love it. Stable, flirty, emotionally tuned—doesn’t overdo the spice but leans into it naturally. |
| Steelskull/L3.3-Mokume-Gane-R1-70b | LLaMA | 16K | Solid generalist with flexible tone and structure | Multi-genre RP, consistent pacing | Can feel plain compared to others in the L3.3 family | 1 | 2 | Less standout than San-Mai, but good for structure-driven storytelling |
| GeneticLemonade/GeneticLemonade-70B | LLaMA | 16K | Smart, stable, minimal repetition; works well across genres | General-purpose RP with consistent logic and flow | Lacks emotional volatility or intense character nuance | 1 | 2 | My go-to when I want something stable, logical, and coherent |
| Sao10K/70B-L3.3-Cirrus-x1 | LLaMA 3.3 | 16K | Stable, thoughtful, lightly poetic tone | Balanced long-form RP, worldbuilding, general character work | May need temp tweaks for sharper emotional spikes | 1 | 2 | A smoother, more stable alternative to Freya. Great baseline model with clean style. |
| allura-org/MN-12b-RP-Ink | Mistral Nemo 12B | 16K | Fast, expressive, high-energy roleplay | Chaotic, dynamic RP with bold tone shifts | Can be overwhelming or unstable without tuned settings | 3 | 2 | Extremely punchy for its size. Called “GOAT” by users. Great for quick and loud character play. Unfiltered vibe. |
| inflatebot/MN-12B-Mag-Mell-R1 | Mistral Nemo 12B | 16K | Rich worldbuilding, vivid prose, strong adaptive memory | Lyrical storytelling, evolving fantasy settings | Can become florid or unstable at high temps | 2 | 2 | Dreamlike prose and structure, built from ‘Best of Nemo’ merges. Surprisingly poetic without fine-tuning. Great for fantasy. |
| Sao10K/70B-L3.3-Cirrus-x1 | LLaMA 3.3 | 16K | Stable, thoughtful, lightly poetic tone | Balanced long-form RP, worldbuilding, general character work | May need temp tweaks for sharper emotional spikes | 1 | 2 | A smoother, more stable alternative to Freya. Great baseline model with clean style. |