Skip to content
VAELICO
✦ The Model Page

Pick your
story engine

Three open models tuned like strange instruments — one cinematic, one precise, one our own. Choose the wolf that matches your world.

Compare ⚡ API quickstart
256Kmax context
NVFP4on B200
Openweights
MODEL GARAGE Vaelico wolf Write / Build / Ship

Our models

Vaelico V1 illustration
★ Flagship
In Development
In-house · Gemma 4 31B

Vaelico V1

Our in-house flagship: Google's Gemma 4 31B, depth-upscaled to 50B and trained on a curated dataset of multi-turn fiction for character consistency and long-form coherence.

Dense · 50B (depth-upscaled from 31B)256K tokensNVFP4Gemma 4 31B (Apache 2.0)1× NVIDIA B200
Live
ON

Artemis

by TheDrummer · Gemma 4 31B

TheDrummer's Artemis 31B — the fan favorite for character roleplay. Expressive, consistent prose with strong character voice, served at near-lossless FP8 precision. Live now in closed beta.

31B Dense64k ContextFP8
In Development
SOON

GLM-Steam 106B

by TheDrummer · GLM-4.5-Air

TheDrummer's GLM-4.5-Air fine-tune — maximum coherence and deep memory across long scenes. Served as a production-ready hosted API.

106B MoE128k ContextNVFP4
In Development
SOON

Iceblink 106B

by zerofata · GLM-4.5-Air

zerofata's GLM-4.5-Air fine-tune — cold, precise and dialogue-forward, optimized for sharp prose and tight pacing.

106B MoE128k ContextNVFP4
Creative freedom Models that follow your story — no corporate guardrails breaking the scene.
Long-context beasts Keep plot threads, character history and lore in the room.
Writer-first Tuned and judged on scene craft, not trivia leaderboards.
Open spirit Open weights, OpenAI-compatible routing, practical deployment.

Compare

Model Best for Context Status
Vaelico V1 Flagship · character consistency, long arcs 256K In dev
GLM-Steam 106B Coherence, deep memory, cinematic prose 128K Launching
Iceblink 106B Sharp prose, dialogue, tone control 128K Launching
B200

API

OpenAI-compatible

curl https://api.vaelico.ai/v1/chat/completions \
  -H "Authorization: Bearer $VAELICO_KEY" \
  -d '{
    "model": "glm-steam-106b",
    "messages": [
      { "role": "user",
        "content": "Open on a rain-soaked harbor at dawn." }
    ]
  }'

Drop-in OpenAI base URL. Works with SillyTavern, Janitor AI, RisuAI & NovelCrafter.