Ready

Engines

Playground

Pick an Engine (DFlash, vLLM, or Transformers), then a model, then Load.

Use vLLM for Hugging Face SafeTensors models. GGUF files stay on DFlash.

Engine: Stopped
API:
Engine stopped. Enable it or load a model.
Engine log
Model Family Scale Source Disk Updated
Loading your model library… Scanning configured folders. Your models will appear here shortly.
Loading models…

Models downloading from Hugging Face appear here with live progress.

Remote inference nodes

Connect other DFlash Console instances on your network. Check health and send a test chat through each node.

Loading nodes…

Loading documentation…

Local AI infrastructure

DFlash Console

A focused control panel for llama-server engines, DFlash speculative decoding, local model libraries, and live runtime diagnostics.

Version v0.3.105 Current Console release
License GNU AGPL v3-or-later Read the license No warranty. You may run, modify, and redistribute covered source under the AGPL terms.
Runtime FastAPI + llama-server Electron shell available for Windows

What this app does

DFlash Console starts and monitors local engine profiles, loads and unloads GGUF models, tracks GPU and token statistics, searches Hugging Face, and exposes OpenAI-compatible local endpoints.

Security and privacy

The Console is designed for one trusted user on one machine. It binds to loopback by default and does not provide multi-user authentication. Do not expose it beyond loopback without adding an authenticated access layer.

Get started

  • Windows installerGitHub Releases (DFlash-Console-Setup-*-x64.exe)
  • Terminal CLIpip install dflash-console then dflash serve
  • Source — clone the repository, copy config.example.json to config.json, run .\run.ps1

Public preview. Report bugs on GitHub Issues. Only one Console API should run on port 8900 at a time.

Desktop runtime

The Windows installer is a thin Electron shell. The backend, documentation, model files, logs, and native llama-server binaries remain in the selected Console data root so large runtime assets are not copied into every installation.

Configure DFLASH_CONSOLE_ROOT or choose the data root when the packaged app first starts.

Attribution

The GNU AGPL requires retaining the copyright and license notices and providing corresponding source when its terms apply. When you redistribute or build on DFlash Console, please also link to the DFlash Console source repository in your README, About page, or project documentation. The DFlash name and logo are governed separately by TRADEMARKS.md.

Support DFlash Console

DFlash Console is free and open source, developed in spare time on local hardware. If it saves you time or helps your projects, a small donation supports development, hardware, and hosting.

Donations go through PayPal and are not required to use any feature of DFlash Console.

Open Model catalog to browse Hugging Face models.

Select a model to view details, README, and download GGUF files.

Downloads & engines

Everything you download after setup: LLM engines (vLLM, Transformers), speech runtimes, and Hugging Face model downloads. Install an engine here, then pick it on the Engines, Playground, or Models tab and load a model.

Engine modules

Extra downloads you install after setup appear first. Speech and GGUF engines bundled with the app are listed below.

Loading components…

Active model downloads

Hugging Face downloads currently in progress

None in progress.

App updates

New Console releases may ship updated speech bundles. Check for signed Windows updates in Desktop app settings.

Desktop app

Tray, startup, and other preferences for the Windows installer. The browser UI ignores these options.

This Console

The browser and the desktop app must show the same address. Both read one config file from this server.

Web address

Open this in a browser to see the same Settings as the desktop app.

Config file

Shared hardware and engine settings

Window & tray

What happens when you close the window or restart your PC.

Start with Windows

Open DFlash Console when you sign in.

Minimize to system tray

Minimizing or closing the window hides it in the tray instead of the taskbar.

Start minimized to tray

When Windows starts DFlash at sign-in, keep the window hidden until you open it from the tray. Opening the desktop shortcut always shows the window.

Show splash on startup

Brief loading window while the local engine starts.

Check for updates automatically

Allow DFlash Console to check periodically. You can always check manually below.

About this install

Version and folders used by the desktop shell.

App version
Console data folder

Backend, models, and config (server.ps1, api, static).

Shell preferences

Tray and startup settings for this install.

Automatic updates

Check the DFlash Console website for signed Windows updates. Downloads are verified before installation.

Not checked

Updates are available only in the desktop app.

Model libraries

Choose where Hugging Face downloads land and which folders the Models tab scans on this PC.

Default download folder

New Hugging Face downloads save here

Download connections

Parallel HTTP ranges for Hugging Face downloads. Files under 32 MiB always use one connection.

Test download speed

Downloads a small sample from Hugging Face and compares 1, 2, 4, 6, and 8 parallel connections on this PC.

Library locations

Enable folders to scan for models on this PC. Use Scan PC to find Piper, Whisper, GGUF, OCR, and other installs automatically.

Paths & backup

Read-only folders used by DFlash Console. Export or import config and launch presets when moving to a new machine.

Config file

Engine profiles, libraries, and preferences

DFlash install

llama-server binaries, draft models, and launch scripts

Model library

Default folder scanned for GGUF models

Engine logs

Boot and load output for each engine profile

Launch presets

Generated .ini files before each router launch

Console UI

Address where this app is served

Import & export

Back up or move your config and generated launch presets.

System

GPUs

Detecting GPUs…

Multi-GPU

Multi-GPU mode

Used when an engine is set to pick the GPU automatically. A GPU you choose by hand on Engines still wins. Changing this reloads loaded models that need a new layout — you do not restart DFlash Console.

Dedicated VRAM only

Prefer dedicated GPU memory over shared memory for weights.

KV cache on GPU

Turn off to keep the active KV cache in system RAM. Disk is used only for saved prompt/slot cache, not live inference.

Live monitor

Refreshes every few seconds while settings stay open.

Engine profiles

Network, runtime, and launch defaults for each engine in config.json. Pick a profile first — changes apply to that engine only.

Engine profile

Which configured engine you are editing

Network & API

OpenAI-compatible endpoint for this profile.

Listen port

API port for this engine

Bind address

127.0.0.1 keeps the API local-only

API endpoint
Console OpenAI gateway

One friendly OpenAI-compatible URL for any client (chat, embeddings, audio)

Gateway port

Single port like LM Studio's 1234 (default 8001). Restart the app after changing.

Default chat engine

Engine that /v1/chat/completions routes to

OpenAI base URL
Runtime

Default limits when this engine loads a model.

Context window (API)

Starting context on load — external API requests may grow it up to the limit below

Context window limit

Hard cap — API requests can never exceed this

Idle unload

Minutes before this Console engine unloads its model when idle · 0 = never · does not affect LM Studio, Ollama, or other external apps

Launch

DFlash preset and GPU override for this profile.

DFlash preset

Target/draft pairing and speculative decoding mode

GPU device

Override automatic GPU selection

Console API reference

Manage engines, load models, and proxy chat completions.

Full guide: open Documentation in the sidebar, or Swagger UI.

API clients

Connect Cursor, Claude Desktop, and other tools to your local engines. MCP hosting is not active yet — use the OpenAI-compatible URLs below.

MCP host status

In-console MCP discovery and registration

Not active
OpenAI-compatible engines

These URLs work today with Playground, API clients, and many integrations

mcp.json preview

Starter config for Cursor, Claude Desktop, or other MCP-aware apps

Speech & runtimes

Device policy, CPU fallback, and memory budget for Piper, Whisper, and other non-llama runtimes. To install vLLM or Transformers, use Downloads & engines.

Install engines

vLLM and Transformers downloads are in one place — not here.

Installed runtimes

Device policy, CPU fallback, and memory budget per runtime

Loading…

GPU contention

Which Console runtimes or external apps are holding VRAM right now

Checking…
Loading behavior

Optional guards when VRAM is busy

Auto-stop other Console runtimes on load

When GPU contention recommends it, unload other Console engines before loading the target

Warn when a runtime runs on CPU

Show a reminder in the Playground when the active TTS/STT runtime is set to CPU

Save runtime settings

Writes runtimes[] back to config.json