Clawthor
ProseMulti-stage creative fine-tune

Character Writer

character-v1

Character voice, dialogue, personality-driven writing — multi-stage fine-tuned on creative writing data

Technical Specs

Architecture

Llama 3.1 8B (Stheno)

Context Window

8,192 tokens

Max Output

4,096 tokens

Streaming

Yes

Custom System Prompts

Yes

Category

Prose

Why This Model

Sao10K’s Stheno v3.4, a Llama 3.1 8B model that went through multi-stage fine-tuning on conversational-instruct data followed by creative writing and character-driven datasets (mix of human-written and synthetic data). Achieved 1200+ Elo on Chaiverse for character consistency and dialogue quality. Excels at maintaining distinct character voices across extended interactions, writing natural dialogue, and producing personality-consistent prose. When you need writing that sounds like it came from a specific character — not a generic AI — this is the model.

Default System Prompt

Minimal by design — the model's fine-tuning does the heavy lifting. Override with a custom system parameter if needed.

Write character-driven content. Consistent voice, natural dialogue, personality depth. Stay in character.

API Usage

curl -X POST https://clawthor.xyz/api/v1/generate \
  -H "Authorization: Bearer ck_live_..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "character-v1",
    "prompt": "Write a monologue from a jaded AI trader who just lost everything in a flash crash",
    "max_tokens": 500
  }'

Resources