Fictor v0.1.0-alpha.3
Pre-releaseFictor v0.1.0-alpha.3
Fictor turns a plain-language description into a small, self-contained browser
app that you can try privately before deciding whether to keep it.
This is an early evaluation release. It is free, carries no success-rate claim,
and does not guarantee that generated applications work. Try results with
examples you already know. The alpha licence requires generated-code review
before reliance or deployment; do that yourself if you can, or obtain
appropriate help.
Published and verified path
- Linux x86-64 is the only build distributed in this release. Its shipped
binary and end-to-end delivery path were exercised on Linux x86-64. - Linux ARM64, macOS, and Windows builds are withheld because they have not
been run on their target systems. - Local Ollama is the default. OpenAI and Anthropic are optional and require
explicit consent before every cloud operation.
The current measured local-model recommendation is qwen2.5-coder:14b, usually
requiring roughly a 12 GB VRAM-class GPU. Other Ollama models can be configured,
but have different quality, speed, memory, and context-window trade-offs.
Privacy and security
The Fictor application has no telemetry, analytics, crash reporting, update
check, account, remote activation, or licence server. On the Build path the
model receives no shell, tools, or filesystem access. Generated applications
run only in Chrome's browser sandbox during Fictor preview and diagnostics.
Read PRIVACY.md, SECURITY.md, and LICENSE inside the archive for the full
terms, actual data flows, limits, and reporting instructions.
Verify the download
Download SHA256SUMS and
fictor_v0.1.0-alpha.3_linux_amd64.tar.gz from this release, then run:
sha256sum -c SHA256SUMSRelease source identity: private Fictor source commit
7f31ab241df700e1c499db5b6a3da50bc099ddf9.