LoginSign Up
← Back to all sparks
N

Novelcrafter

COLLAB
Velocity0.0

Writing platform for novelists with a story codex and AI drafting tools

www.novelcrafter.com ↗

Novelcrafter keeps the model layer swappable and spends its own effort on the manuscript instead.

ai-writingmodel-routingcodex-contextfiction-toolsexports
◆Current state
Two threads run through this window. The first is model plumbing: default collections refreshed as Claude 4.5 Sonnet, Gemini 3 Pro, and GPT-5.1 arrive, a generic OpenAI-compatible provider so unsupported vendors work anyway, and an opt-in control to prefer or avoid reasoning across Scene Beats and Chats. The second is the writer's own data — Codex entries can be marked case-sensitive so a character named Will stops matching every sentence, and an entry can sit in several custom categories at once. Word export gained bullet and numbered lists. The feed stops in March 2026.
◆Where it's heading
The model layer is treated as commodity and kept deliberately replaceable: users can clone collections, revert to previous ones, or point at any OpenAI-compatible endpoint, and the reasoning control is a preference rather than a hardcoded behavior. Novelcrafter's own investment goes into the Codex — the structured record of characters, places, and rules that gets fed to whichever model is in play — and into getting manuscripts out cleanly. That is a defensible split for a tool whose value is the writer's accumulated context rather than the generation itself.
◆Prediction
Expect further Codex precision work, since case-sensitivity and multi-category membership are both fixes for the same failure — the wrong entries being pulled into context. What the entries do not show is whether the experimental reasoning support graduates out of the settings flag.

◆Recent moves

  1. 5mo ago

    New Codex Tracking Options

    Codex entries can be marked case-sensitive and can belong to more than one custom category at a time, both promoted out of experimental. It fixes over-matching on common words like Don or Will, which is a context-quality problem rather than a cosmetic one — the Codex decides what the model sees.

    View source ↗
  2. 7mo ago

    List support for Word export

    Bullet and numbered lists survive Word export, alongside a reworked help popup, unread-chat banners, and a system status indicator. Export fidelity is the last mile for a drafting tool, and this is one more format that no longer needs fixing by hand.

    View source ↗
  3. 7mo ago

    January 9, 2026 - AI Thinking support

    Reasoning support ships to everyone as an experimental opt-in, exposed as a general prefer-or-avoid choice across Scene Beats and Chats with depth tuned through custom model parameters. The generic framing is the notable part — it works across models that reason differently rather than binding to one vendor's implementation.

    View source ↗
  4. 9mo ago

    Gemini 3 Pro joins the default model collections

    Gemini 3 Pro joins the system model collection and GPT-5 moves to GPT-5.1, with the usual documented path back to the previous collections. Routine model-catalog maintenance in a product built to keep that layer interchangeable.

    View source ↗
  5. 10mo ago

    Default model collections add Claude 4.5 Sonnet

    Default collections refresh to include Claude 4.5 Sonnet and newer versions of existing models, dropping low-usage entries. Custom collections are untouched, which is the point of keeping them user-owned.

    View source ↗
  6. 11mo ago

    Generic OpenAI-compatible provider opens up AI connections

    A generic OpenAI-compatible provider means niche vendors without official support can be connected anyway, alongside broader rework of the AI connection internals. It is the clearest statement of the strategy — the model layer stays open rather than curated.

    View source ↗