Skip to content

The native Lumiverse version of DEUS EX MACHINA.

DEUS EX MACHINA (DEM) is a feature-rich, beginner-friendly, contextually dynamic, truly modular preset focused on collaborative story writing.

Published November 18, 2026

views
27
downloads
DEUS EX MACHINA V4 Lumiverse

Adult · blurred

DEUS EX MACHINA V4 Lumiverse

PresetAdult content

Lumiverse preset · works in Lumiverse

v1.0 · November 18, 2026

How to use this

There is a README inside the preset explaining the model setup. It's in the prompt list. You can read it for specific instructions regarding the different models.

The default settings of this preset REQUIRE reasoning to be either disabled or set to the lowest effort possible!

Recommended models for this preset: Opus 4.6 (smartest; expensive; detailed prose); Gemini 3.7 Flash (smart; cheap; realistic characters); GLM 5.3 (smart; cheap; natural dialogue); DeepSeek V4 Pro 0813 (smart; cheap; balanced qualities); Gemma 4 31b (not very smart, but follows instructions really well; very cheap).

Prompt fragments

173 fragments · 166 switched on

  1. ⬆️ CONFIGURE PRESET ⬆️

    System · Before the conversation

    {{// Open Configure Prompt Variables, the button above this list, to enable and disable modules and pick options. Keep every block enabled; the settings there decide what each one sends}}{{trim}}

  2. ⚠️ README Model Setup ⚠️

    System · Before the conversation

    {{//

    Set this preset up in Configure Prompt Variables: the button at the top of the Loom block list (Preset tab). Keep the blocks enabled; the settings there decide what each one sends.

    IMPORTANT: The default settings of this preset REQUIRE reasoning to be either disabled or set to the lowest effort possible! Exceptions: GLM 5.3 (reasoning always on, low; keep Planning on Scene Plan) and Gemini 4 Argon -- set Planning to Thinking (Fallback) in Configure Prompt Variables

    Recommended models for this preset: Opus 4.6 (smartest; expensive; detailed prose); Gemini 3.7 Flash (smart; cheap; realistic characters); GLM 5.2/3 (smart; cheap; natural dialogue); DeepSeek V4 Pro 0813 (smart; cheap; balanced qualities); Gemma 4 31b (not very smart, but follows instructions really well; very cheap).

    Samplers: in Lumiverse a sampler is off only when empty or 0; a 1 is still sent, and providers without it (e.g., OpenAI Responses) reject the request with "Unsupported request field". Leave Top K unchecked. If the error names a sampler you cannot see, press Reset in the sampler panel and re-enter temperature, Top P and Max Response

    Most models (that support disabling reasoning): Claude Opus 4.6, Kimi K2.5/2.6, Mimo V2.5 Pro/2.6, DeepSeek V4 Pro, DeepSeek V4.1 Flash, DeepSeek V3.2, GLM 4.7, Gemma 4 31b, etc.

    • Samplers: temperature - 0.75-1.0 (Exceptions: Gemma 4 1.0, Opus 1.0. For the rest, start with 0.75, the default); Top P - 0.95 (default); rest - 1.0 or disabled (default; exceptions: Gemma 4 31b - Top K: 64)

    • API Reasoning: on / checked

    • Reasoning effort: None

    • Planning: Scene Plan
      Setup: Reasoning Effort: None is what turns reasoning off, and Lumiverse sends it only while "API Reasoning" is checked. If your provider has no None option, set Reasoning Effort to Auto, then in the Reasoning tab -> Reasoning / CoT -> "Custom Body", tick Enabled and enter: {"reasoning": {"effort": "none"} }
      If on NanoGPT: Choose a non-thinking variant of the model, e.g., DeepSeek V4 Pro 0813 instead of DeepSeek V4 Pro 0813 Thinking

    GLM 5.2

    • API Reasoning: on / checked

    • Reasoning effort: None

    • Planning: Scene Plan
      Provider-specific settings: the same as in the "most models" section above

    GLM 5.3 (Reasoning always on)

    • Samplers: temperature - 0.75 (default); Top P - 0.95 (default); rest - 1.0 or disabled (default)

    • API Reasoning: on / checked

    • Reasoning effort: low

    • NanoGPT model: GLM 5.3; GLM 5.3 Thinking also always reasons but ignores reasoning effort, so it overthinks

    • Planning: Scene Plan

    GLM 5.3 Flash (Reasoning always on)

    • Samplers: temperature - 0.75 (default); Top P - 0.95 (default); rest - 1.0 or disabled (default)

    • API Reasoning: on / checked

    • Reasoning effort: low

    • Planning: Scene Plan

    Kimi K3 (Reasoning always on)

    • Samplers: temperature - 1.0; Top P - 0.95 (default); rest - 1.0 or disabled (default)

    • API Reasoning: on / checked

    • Reasoning effort: low

    • Planning: Scene Plan

    Gemini 3.1 Pro, 3.5-3.8 Flash (Reasoning always on)

    • Samplers: temperature - 1.0; Top P - 0.95 (default); rest - 1.0 or disabled (default)

    • API Reasoning: on / checked

    • Reasoning effort: low; for other OpenAI-compatible providers, set it to Auto and enter {"thinkingLevel": "low"} in Custom Body, Enabled ticked

    • Planning: Scene Plan

    • Passthrough: Gemini, if replies are blocked by the filter

    Gemini 4 Argon (Reasoning always on)

    • Provider: Google Vertex (recommended)

    • Samplers: ignored by Argon (temperature, Top P, Top K have no effect); leave the defaults

    • API Reasoning: on / checked

    • Reasoning effort: low; for other OpenAI-compatible providers, set it to Auto and enter {"thinkingLevel": "low"} in Custom Body, Enabled ticked. Never None or minimal: Argon rejects them with an error

    • Planning: Thinking (Fallback) (Configure Prompt Variables: REASONING → Planning)

    • Content Policy and Phi's Vow: off (Configure Prompt Variables: CORE)

    • Passthrough: Gemini, on

    }}{{trim}}

  3. 🔹CORE🔹4 fragments
  4. ⚠️ Macro Cleaner ⚠️

    System · Before the conversation

    {{.director_state = }}
    {{.cyoa = }}
    {{.smut = }}
    {{.smut_mxl = }}
    {{.gore = }}
    {{.gore_ultra = }}
    {{.improve_input = }}
    {{.character_realism = }}
    {{.anti_repetition = }}
    {{.banned_terms = }}
    {{.anti_positivity = }}
    {{.dark_portrayals = }}
    {{.anti_omniscience = }}
    {{.story_threads = }}
    {{.threads_ledger = }}
    {{.tension = }}
    {{.pacing = }}
    {{.combat_writing = }}
    {{.world_building = }}
    {{.fandom = }}
    {{.fandom_name = }}
    {{.momentum_engine = }}
    {{.undercurrents = }}
    {{.relationships = }}
    {{.dialogue_colors = }}
    {{.expressive_dialogue = }}
    {{.drawn_out_words = }}
    {{.stammers = }}
    {{.caps_volume = }}
    {{.vocal_sounds = }}
    {{.dialogue_sounds = }}
    {{.visual_storytelling = }}
    {{.narrative_styles = }}
    {{.narrative_comedic = }}
    {{.narrative_dramatic = }}
    {{.narrative_epic = }}
    {{.narrative_gothic = }}
    {{.narrative_dreamy = }}
    {{.narrative_eerie = }}
    {{.narrative_surrealist = }}
    {{.narrative_grimdark = }}
    {{.narrative_lighthearted = }}
    {{.narrative_intimate = }}
    {{.narrative_suspenseful = }}
    {{.narrative_melancholic = }}
    {{.narrative_styles_checkpoint = }}
    {{.status = }}
    {{.psychological_states = }}
    {{.plotlines = }}
    {{.fate = }}
    {{.diaries = }}
    {{.meanwhile = }}
    {{.hidden_truths = }}
    {{.new_faces = }}
    {{.companions = }}
    {{.companion_chi = }}
    {{.companion_lou = }}
    {{.companion_sam = }}
    {{.companion_grace = }}
    {{.companion_layla = }}
    {{.companion_phi = }}
    {{.companion_sola = }}
    {{.companion_friend = }}
    {{.companion_friends = }}
    {{.companion_no_mode = }}
    {{.scratchpad = }}
    {{.tracker = }}
    {{.language = }}
    {{.true_thoughts = }}
    {{.true_thoughts_visible = }}
    {{.pick_agency = }}
    {{.pick_tension = }}
    {{.pick_status = }}
    {{.pick_relationships = }}
    {{.pick_companions = }}
    {{.pick_pacing = }}
    {{.pick_combat_writing = }}
    {{.pick_world_building = }}
    {{.pick_narration = }}
    {{.pick_dialogue = }}
    {{.pick_nsfw = }}
    {{.pick_violence = }}
    {{.pick_length = }}
    {{.pick_thoughts = }}
    {{.pick_reasoning = }}
    {{.pick_tense = }}
    {{.pick_pov = }}
    {{.user_first_person_pov_invalid = }}
    {{.pick_char_pov = }}
    {{.pick_asterisks = }}
    {{.main_statements = }}
    {{.response_directives = }}
    {{.momentum_engine_router = }}
    {{.fate_outcomes = }}
    {{.scene_plan = }}
    {{.thinking = }}
    {{.anti_overthink = }}
    {{.agency_order = }}
    {{.character_realism_checkpoint = }}
    {{.sheet_open = }}
    {{.sheet_closed = }}
    {{.post_history = }}
    {{.persona_open = }}
    {{.persona_closed = }}
    {{.story_development_open = }}
    {{.story_development_closed = }}
    {{.response_tracker = }}
    {{.response_cyoa = }}
    {{.response_true_thoughts = }}
    {{.response_status = }}
    {{.response_story_threads = }}
    {{.response_psychological_states = }}
    {{.response_plotlines = }}
    {{.response_diaries = }}
    {{.response_meanwhile = }}
    {{.response_hidden_truths = }}
    {{.response_new_faces = }}
    {{.response_companions = }}
    {{.response_scratchpad = }}
    {{.response_tension = }}
    {{.response_relationships = }}
    {{.response_momentum_engine = }}
    {{.order_error_core = }}
    {{.order_error_xml = }}
    {{.order_error_agency = }}
    {{.order_error_nsfw = }}
    {{.order_error_tension = }}
    {{.order_error_relationships = }}
    {{.order_error_narrative_styles = }}
    {{.order_error_response_directives = }}
    {{.order_error_main_statements = }}
    {{.order_error_final = }}
    {{.preset_error = }}
    {{.format_missed = }}
    {{.dem_ooc = }}{{#if {{matches::{{lastUserMessage}}::^(?=[\s\S]*?(?:\bOOC\b|((|\bout[ -]of[ -]character\b|\bPhi\b))::i}}}}{{.dem_ooc = on}}{{/if}}
    {{.braille_blanks = }}
    {{.macro_cleaner = on}}{{trim}}

Unsupported formatting is hidden here. The original text is preserved in downloads.

Settings

25 settings

Samplers9 settings

  • Temperature
    0.75

    Higher values wander further from the model's likeliest next word.

  • Top P
    0.95

    Keeps the likeliest words until their odds add up to this much.

  • Min P
    0

    Drops words whose odds fall this far under the best one's.

  • Maximum reply length
    10,000

    The longest reply the model is allowed to write.

  • Context size
    1,000,000

    How much of the conversation is sent with each reply.

  • Frequency penalty
    0

    Pushes harder the more often a word has been used.

Completion12 settings

  • Send a system prompt
    No
  • Merge system messages
    Yes

    A run of system messages is sent as one.

  • Character names in messages
    0

    How a speaker's name is attached to what they say.

  • Assistant prefill
    Nothing

    Text the model's reply is made to start with.

  • Impersonation prefill
    Nothing

    Text a reply written as the user is made to start with.

  • Reasoning prefill
    Nothing

    Text the model's reasoning is made to start with.

Advanced4 settings

  • Seed
    -1

    The same seed and the same request give the same reply.

  • Stop strings

    The model stops writing when it reaches one of these.

  • Collapse consecutive messages
    No
  • Trim a cut-off word
    No

Regex scripts

183 scripts

  • Restore spaces from Braille blanks in saved reply

    (?<!\ufffd)\ufffd*\u2800

  • Restore spaces from Braille stand-ins in saved reply

    {{#if .braille_blanks}}{{else}}(?!){{/if}}(?:(?<!�)�+[  - ␠␣]|[ - ␠␣]|(?<=[^\s═])═+(?=[^\s═]))

  • Restore spaces from Braille blanks

    (?<!\ufffd)\ufffd*\u2800

  • Hide @Tracker scene backdrop (enable for settings it cannot fit)Off

    (?=<tracker\b)<style>.dem-track .dem-track__grid{background:var(--dem-paper)!important;padding-top:10px!important}.dem-track .dem-track__grid:before,.dem-track .dem-track__grid:after,.dem-track .dem-track__scene{display:none!important}</style>

Nudges

7 nudges

  1. Start of a chat

    [Start a new story]
  2. Start of a group chat

    [Start a new group chat. Group members: {{group}}]
  3. Continue a cut-off reply

    [Continue the following message. Include NONE of the original message. Use capitalization and punctuation as if your reply is a part of the original message: {{lastMessage}}]