Qwen3.8-Flash-Next on any consumer hardware: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.
Xolotl (formerly Strata-DSV4.1): DeepSeek-V4.1-Flash, all 552B, on one gaming PC (RTX 4090 + 90 GB RAM) at ~27 tok/s, plus GLM-5.3-Flash, Qwen3.8-Flash-Next and Qwen3.8-27B. A heavily extended fork of Strata: MoE experts tiered across VRAM, pinned RAM and NVMe; OpenAI + Anthropic API.
An open-source file manager for Android: two panes, archives, APK and DEX editing, a code editor, network storages and an encrypted vault. No ads, no accounts, no tracking.
Run large AI models on your own PC: Qwen3.8-Flash-Next on Strata's engine, and GLM 5.3 Flash, DeepSeek V4 Flash, LFM2.5, Qwen3.6, Ornith 1.5, Gemma 4, Tiel Coder and MiniCPM-V on a second engine written for PolyStrata. One-click install for Windows and Linux, OpenAI/Anthropic API on localhost.
The best file manager on Linux
Strata minus the bullshit. clone, rsync to prod, compile and go
Qwen3.8-Flash-Next (125B MoE) 单卡 16G 本地部署实测:Strata + GSQ-RCO IQ3_XXS,Ubuntu 24.04
存目 — 以你为主角的个人百科 / Strata — a personal encyclopedia about you
The knowledge compiler. Raw sources in, interlinked wiki out. Inspired by Karpathy's LLM Wiki pattern.
Strata Manager makes Niko1221/Strata, the local LLM inference engine, easier to set up and use. Manage local models and services, monitor hardware and API usage, chat and tune inference. Open-source desktop app for Windows / Linux x64.
Strata fork: Qwen3.8-Flash-Next 125B MoE in NVFP4 on one RTX 20-50 card (12 GB+) and 64 GB of RAM or more. Setup installs our GPTQ quants (huihui abliterated, OrcaRouter) from Hugging Face. CPU experts on AVX-512 or AVX2 (+AVX-VNNI), all experts in RAM from 92 GB, a low-RAM mode below, vision, W4A8 prompts. Windows / CUDA 13.
Strata Files: a fast, native, keyboard-first file manager and Finder alternative for macOS. Instant search, live folder sizes, disk usage map, dual pane.
Open source analytics and market risk library from OpenGamma
The Stratagus strategy game engine
Qwen3.8-Flash-Next on Apple Silicon: an MLX engine that reads Strata's GGUF files and speaks its line protocol (unofficial)
Ultimate assistant for Helldivers 2. Features: God Mode, Infinite Ammo, Infinite Grenades, Infinite Stims, No Reload, Unlimited Stratagems, Super Credits & Medals Farm, Infinite Mission Timer, and more. 20+ functions.
Real-time wgsl visualisation tooling for educating oneself in the art of shader programming
Think before doing, prove before saving. Vestige is a local, deterministic safety kernel for AI agents. Every memory write and agent action is recorded in Strata, an append-only log, so you can see exactly what your agent did and why.
Stratawright is an agentic DAW that your AI agent (Claude Code, Cursor, Codex) can natively control so you can focus purely on making music.
Session strata for the DeepSeek Harness Web GUI: the transcript's scrollbar becomes a to-scale, colour-coded map of the whole run, with your own messages emphasised and clickable anchors.
UNCANNY is a real-time remaster engine for PC games and emulators with STRATA reconstruction, Motion Guard, REVENANT geometry/material upgrades, water recovery, optional DLSS5 and multi-API support.
Read-only dashboard and rack panel for NVIDIA DGX Spark-class nodes, other NVIDIA GPU Linux machines and Apple Silicon Macs running vLLM, SGLang, TensorFold, llama.cpp, Strata or oMLX
In-game stat editor for Helldivers 2: weapons, beams, stratagems and sentries. Press F8, change the numbers, saved for every launch.
Helldivers 2 Companion Loadout Planner 2026: Super Samples, Medals and Stratagem Tracker
Helldivers 2 Solo Trainer 2026: God Mode, Infinite Ammo & Stratagems
Helldivers 2 Stratagem Hero Trainer & Arcade Companion 2026
Stratagem Hero 2026 Loadout Planner and Trainer for Helldivers 2 Arcade
Semantic modding runtime and guarded API for Helldivers 2, with mapped weapons, stratagems, vehicles, boosters, backpacks, and more.
A better stratagem bounce mod for Helldivers 2.
Strata fork that runs GLM-5.3-Flash (320B-A18B, 157 GB) on one PC: RTX 5090 + 128 GB RAM + NVMe, ~30 tok/s chat. Fork of Niko1221/Strata (MIT).
See why a web vulnerability works, not just how to trigger it — a single-file, zero-dependency security lab that renders a vulnerable app as a 7-layer cross-section. 17 labs, attacker & defender modes, AI-authorable.
在 8 GB 显存的显卡上跑 35B MoE(Cyber-Tiel-Coder-35B-A3B):专家权重流式上卡、未命中专家由 CPU 池就地计算。Strata 引擎的 fork(MIT)。RTX 4080S 实测 8 GB 档 57.7–63.5 tok/s、全常驻 134–136 tok/s。Build from THIS repo — upstream does not support this model.