{
    "archive_path": "archive/1791151586.920494",
    "base_url": "github.com/Niko1221/Strata",
    "basename": "Strata",
    "bookmarked_date": "2026-10-04 22:06",
    "canonical": {
        "archive_org_path": "https://web.archive.org/web/github.com/Niko1221/Strata",
        "dom_path": "output.html",
        "favicon_path": "favicon.ico",
        "git_path": "git/",
        "google_favicon_path": "https://www.google.com/s2/favicons?domain=github.com",
        "headers_path": "headers.json",
        "htmltotext_path": "htmltotext.txt",
        "index_path": "index.html",
        "media_path": "media/",
        "mercury_path": "mercury/content.html",
        "pdf_path": "output.pdf",
        "readability_path": "readability/content.html",
        "screenshot_path": "screenshot.png",
        "singlefile_path": "singlefile.html",
        "warc_path": "warc/",
        "wget_path": "github.com/Niko1221/Strata.html"
    },
    "domain": "github.com",
    "extension": "",
    "hash": "HGP7KS2NTN5MAX34T4EE",
    "history": {
        "archive_org": [],
        "dom": [],
        "favicon": [
            {
                "cmd": [
                    "curl",
                    "--silent",
                    "--location",
                    "--compressed",
                    "--max-time",
                    "120",
                    "--output",
                    "favicon.ico",
                    "--user-agent",
                    "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/118.0.0.0 Safari/537.36 ArchiveBox/0.7.4 (+https://github.com/ArchiveBox/ArchiveBox/) curl/curl 8.14.1 (x86_64-pc-linux-gnu)",
                    "https://www.google.com/s2/favicons?domain=github.com"
                ],
                "cmd_version": "curl 8.14.1 (x86_64-pc-linux-gnu)",
                "end_ts": "2026-10-04T22:10:07.041020+00:00",
                "index_texts": null,
                "output": "favicon.ico",
                "pwd": "/data/archive/1791151586.920494",
                "schema": "ArchiveResult",
                "start_ts": "2026-10-04T22:10:06.459749+00:00",
                "status": "succeeded"
            }
        ],
        "git": [],
        "headers": [
            {
                "cmd": [
                    "curl",
                    "--silent",
                    "--location",
                    "--compressed",
                    "--head",
                    "--max-time",
                    "120",
                    "--user-agent",
                    "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/118.0.0.0 Safari/537.36 ArchiveBox/0.7.4 (+https://github.com/ArchiveBox/ArchiveBox/) curl/curl 8.14.1 (x86_64-pc-linux-gnu)",
                    "https://github.com/Niko1221/Strata"
                ],
                "cmd_version": "curl 8.14.1 (x86_64-pc-linux-gnu)",
                "end_ts": "2026-10-04T22:10:07.741817+00:00",
                "index_texts": null,
                "output": "headers.json",
                "pwd": "/data/archive/1791151586.920494",
                "schema": "ArchiveResult",
                "start_ts": "2026-10-04T22:10:07.046709+00:00",
                "status": "succeeded"
            }
        ],
        "htmltotext": [],
        "media": [],
        "mercury": [
            {
                "cmd": [
                    "/app/node_modules/@postlight/parser/cli.js",
                    "https://github.com/Niko1221/Strata"
                ],
                "cmd_version": "1.0.0",
                "end_ts": "2026-10-04T22:10:29.416556+00:00",
                "index_texts": null,
                "output": "mercury",
                "pwd": "/data/archive/1791151586.920494",
                "schema": "ArchiveResult",
                "start_ts": "2026-10-04T22:10:23.719744+00:00",
                "status": "succeeded"
            }
        ],
        "pdf": [],
        "readability": [
            {
                "cmd": [
                    "/app/node_modules/readability-extractor/readability-extractor",
                    "/tmp/tmpca1xkw8w",
                    "https://github.com/Niko1221/Strata"
                ],
                "cmd_version": "0.0.11",
                "end_ts": "2026-10-04T22:10:23.706553+00:00",
                "index_texts": [
                    "Strata\nEnglish \u00b7 \u7b80\u4f53\u4e2d\u6587 \u00b7 \u65e5\u672c\u8a9e \u00b7 Deutsch \u00b7 Fran\u00e7ais \u00b7 Espa\u00f1ol \u00b7 Portugu\u00eas\nRun a 125-billion-parameter AI model on your own gaming PC\nNVIDIA or AMD graphics card (12 GB or more) \u00b7 Windows or Linux \u00b7 free and open source\n\nA voxel pagoda garden, 1 shot prompt running on an RTX 5070 with Strata (IQ3_S, 128K context) \u00b7\nfull video (49 s)\nStrata runs Qwen3.8-Flash-Next on a normal PC. This is a\nlarge, smart AI model that usually needs a server. It chats, writes code, reads pictures and works with your apps\nand coding agents. Nothing leaves your PC.\nHow fast is it?\nWe measured it on two ordinary gaming PCs. A token is about \u00be of a word.\n\nWrites answers: how fast the reply appears in a short chat. 60 tokens per second is faster than you can read.\nReads your prompt: how fast it takes in what you send (here a 32K-token document, code or chat history).\n\n\nNVIDIA: RTX 5070 (12 GB), Ryzen 5 7600, 64 GB RAMAMD: RX 9070 XT (16 GB), Ryzen 9 3900X, 47 GB RAM\n\n\n\n\nSize\nWrites answers\nReads your prompt\n\n\n\n\nQ2_0\n94 tokens/s\n2,650 tokens/s\n\n\nIQ2_XS\n79 tokens/s\n2,090 tokens/s\n\n\nIQ3_XXS\n62 tokens/s\n1,750 tokens/s\n\n\nIQ3_S\n53 tokens/s\n1,620 tokens/s\n\n\nCoder\n55 tokens/s\n2,180 tokens/s\n\n\n\n\n\n\n\nSize\nWrites answers\nReads your prompt\n\n\n\n\nQ2_0\n60 tokens/s\n1,160 tokens/s\n\n\nIQ2_XS\n52 tokens/s\n1,110 tokens/s\n\n\nCoder\n44 tokens/s\n1,420 tokens/s\n\n\n\n\n\nNVIDIA: Q2_0 with engine 0.1.36, the other rows with 0.1.26 (4K answers, 32K prompts). The full tables are in\nDETAILS.md. A card with more VRAM is faster: an RTX 3090 (24 GB) should write\nabout 100-140 tokens per second. Long chats and other cards: speed of each model,\ncommunity results.\n\nStrata is free. If it runs well on your PC, a coffee keeps the work on it going.\nWhat you need\n\n\n\n\n\n\n\n\n\nGraphics card\nNVIDIA GeForce RTX 20, 30, 40 or 50 series, or AMD Radeon RX 7900 XT / XTX, RX 7800 XT / 7700 XT, RX 9060 XT, RX 9070 / 9070 XT, Radeon AI PRO R9700 or RX 6800 / 6900 series. It needs 12 GB of VRAM or more.\n\n\nRAM\n32 GB or more. Your RAM decides which model fits. 64 GB runs every size.\n\n\nDisk\nAbout 80 GB free. Use an SSD if you can: the first start is much faster.\n\n\nSystem\nWindows 10 / 11 or Linux, and a current graphics driver from NVIDIA or AMD.\n\n\n\nThe installer sets up everything else. Two or three cards can share the model (multi-GPU).\nExperimental, written and tested by community members on their own machines:\n\nOlder graphics cards (Tesla P40 / V100, GTX 10, Radeon VII / MI50, RX 6700 XT, RX 5500 XT): Older GPUs.\nIntel Arc, built from source on Linux: Intel Arc.\nOlder processors without AVX2: they work, but slowly. Older CPUs.\n\nThe full list: docs/INSTALL.md.\nInstall\nLet your AI set it up\nDo you use an AI coding assistant (Claude Code, Cursor, Codex, GitHub Copilot, ...)? Paste this into it:\nSet up Strata on this PC for me: https://github.com/Niko1221/Strata - follow docs/AI_SETUP.md in that repository.\n\nIt checks your graphics card, RAM and disk and picks the model that fits. Then it installs and starts it and tells\nyou how to connect your apps. AI tools can also install, start and stop Strata through its\nMCP server.\nOr do it yourself\nDownload Strata and unzip it (or git clone it).\nWindows: double-click START-HERE.bat. Linux: run ./setup.sh in the Strata folder.\nThe steps are the same for NVIDIA and AMD. The installer finds your card and sets up the right engine for it. It\nasks you a few questions:\n\nwhich model and which size,\nhow much context (how much text the model keeps in mind),\nwhether it should read pictures.\n\nPress Enter each time for the recommended answer. Then it downloads the model (about 70 GB) and starts it. If the\ndownload stops, run it again: it continues where it left off. Your browser opens the Strata app at\nhttp://127.0.0.1:8080.\n\nWhile the model starts, your PC can be slow or stop responding for 1-3 minutes (longest the first time).\nStrata loads 35-55 GB into your RAM and locks part of it for the graphics card. This is normal. Wait, and don't\nclose the window. The window shows what Strata is doing.\n\nNext time, run START-HERE.bat (or ./setup.sh) again. It starts right away and downloads nothing twice. Close\nits window to stop the model. UPDATE.bat (./update.sh) updates Strata without starting it. Updating, Docker,\nseveral cards, where the files go and every option: docs/INSTALL.md.\nWhich model should I pick?\nThe installer recommends one for your RAM. The same model comes in several sizes, compressed more or less. Smaller\nsizes are faster. Larger sizes are a bit smarter.\n\n\n\nYour RAM\nTake\nWhy\n\n\n\n\n32 GB\nCoder\nit fits 32 GB, and it is made for code (with a 24 GB card, Q2_0 and IQ2_XS run too)\n\n\n48 GB\nIQ2_XS (or Q2_0, the fastest)\nthe larger sizes do not fit\n\n\n64 GB\nIQ2_XS (recommended), or IQ3_XXS / IQ3_S\nevery size fits; IQ3_S is the best and the slowest\n\n\n96 GB or more\nIQ3_S, or Unsloth's UD-IQ4_XS (~4-bit)\nroom for the largest sizes with everything else open\n\n\n\n\nCoder: a coding version with half of the experts removed. It reaches 91% of the full\nmodel's SWE-bench Verified score (measured by its authors) and fits 32 GB of RAM. It is weaker outside code,\nincluding Chinese and other CJK text (#438). For those, take Q2_0, IQ2_XS or IQ3_S, which keep every expert.\nSwift 1.5: a fine-tune that thinks for a much shorter time before it answers. You\nget the answer sooner, at about the same quality.\nUnsloth UD-IQ4_XS: Unsloth's ~4-bit version, between IQ3_S and\nUD-Q4_K_XL in quality. A 94 GB download. With less than ~80 GB of RAM, Strata reads part of it from the SSD\nwhile it answers, so it is slower there (an NVMe SSD helps).\nUnsloth UD-Q4_K_XL (experimental): the closest to the full\nmodel. But Strata reads most of it from the SSD while it answers, so it writes only 7-8.5 tokens/s on a 64 GB PC.\nOrcaRouter's Uncensored IQ3_XXS: you set it up by hand. It is\nnot in the installer's menu.\n\nSizes, downloads and what fits where: docs/MODELS.md. To add another model later, run\nSETUP.bat (Linux: ./setup.sh --setup).\nUsing it\n\nThe Strata app's Monitor (left) while a coding agent writes the pagoda garden from the video (right)\n\nIn the browser: open http://127.0.0.1:8080. It has Chat, a live Monitor of the model and your\nGPU/CPU/RAM, and About with the settings and addresses.\nYour apps and coding agents: add an \"OpenAI-compatible\" provider with the base URL\nhttp://127.0.0.1:8080/v1. Any API key and any model name work.\n\nApps that use Anthropic's API: http://127.0.0.1:8080/v1/messages (Claude Code:\nANTHROPIC_BASE_URL=http://127.0.0.1:8080).\nCodex CLI and other apps that use the OpenAI Responses API: /v1/responses\n(setup).\n\n\nThinking: choose off, low, medium or high in the chat menu or in your app's \"reasoning effort\". Off is the\nfastest. High is best for hard questions.\nPictures: say yes to \"Images?\" in setup. Then click Picture in the chat, or attach pictures in your app.\nAMD cards read pictures on Linux through the processor; on Windows they can't yet.\nFrom your phone or another PC: START-HERE.bat --setup --host 0.0.0.0 --api-key <secret>. Always set a key.\nOne request at a time: by default Strata answers one request, and the others wait. To answer several at once,\nset \"parallel\": 2 (BATCHING.md). On a 12 GB card this makes each answer slower.\nLong prompts: Strata reads the first message of a chat in full, about 1 minute per 30,000 tokens. Follow-up\nmessages start in seconds.\n\nMore: where your chats are stored, the API.\nSomething went wrong?\n\nMy PC froze the first time Strata started. This is normal while it loads the model. Wait, and don't close the\nwindow. Still frozen after 10 minutes? Restart the PC, close other programs and try again, or pick a smaller size.\nIt stopped while downloading or installing. Run START-HERE.bat (or ./setup.sh) again. It continues where\nit stopped.\nIt's very slow and the disk light keeps blinking, or it says \"the engine stopped unexpectedly\". Your PC does\nnot have enough free RAM. Close other programs (browsers use a lot), or pick a smaller size (Q2_0 or IQ2_XS).\nIt says port 8080 is already in use. Strata is already running. Look for its window.\n\nMore problems and their fixes: docs/TROUBLESHOOTING.md. Still stuck? Open an\nissue and attach strata-<model>.log from the Strata folder. Found a\nsecurity problem? Report it privately: SECURITY.md.\nHow does it work?\nModels like this one usually run on servers with hundreds of gigabytes of graphics memory. Your graphics card has\n12-24 GB. Strata makes the model fit by sharing the work across your whole PC. Think of a kitchen: the things\nyou use all the time stay on the counter, and the rest waits in the pantry.\n\n\nThe model is a team of 24,576 small specialists (\"experts\"). Each word needs only 10 of them.\nYour graphics card keeps the few thousand experts that are used most often. Your RAM holds all of them,\nand your processor works on the rest at the same time. Your SSD holds a big lookup table.\n\n\n\nGuess, then check: a small helper guesses the next few words. The big model checks them all at once. You get\nthe same answer, 1.6-1.8x sooner.\nLong texts are read in big pieces (up to 8,192 tokens at a time), at over 1,000 tokens per second.\n\nThe longer explanation: docs/HOW_IT_WORKS.md. Every part and its numbers:\nthe details and the paper.\nCredits and license\nThe model is Qwen3.8-Flash-Next by the Qwen team. It was\ncompressed by ISTA-DASLab, UkisAI (Swift 1.5)\nand Unsloth. Strata uses parts of llama.cpp / ggml. All credits:\ndocs/HOW_IT_WORKS.md. Strata is open source under the MIT License. A few\nparts and every model have their own licenses (which ones).\nSupport Strata\nStrata is free and open source. If it is useful to you, you can support its development:"
                ],
                "output": "readability",
                "pwd": "/data/archive/1791151586.920494",
                "schema": "ArchiveResult",
                "start_ts": "2026-10-04T22:10:19.271856+00:00",
                "status": "succeeded"
            }
        ],
        "screenshot": [],
        "singlefile": [],
        "title": [
            {
                "cmd": [
                    "curl",
                    "--silent",
                    "--location",
                    "--compressed",
                    "--max-time",
                    "120",
                    "--user-agent",
                    "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/118.0.0.0 Safari/537.36 ArchiveBox/0.7.4 (+https://github.com/ArchiveBox/ArchiveBox/) curl/curl 8.14.1 (x86_64-pc-linux-gnu)",
                    "https://github.com/Niko1221/Strata"
                ],
                "cmd_version": "curl 8.14.1 (x86_64-pc-linux-gnu)",
                "end_ts": "2026-10-04T22:10:19.267363+00:00",
                "index_texts": null,
                "output": "GitHub - Niko1221/Strata: 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. \u00b7 GitHub",
                "pwd": "/data/archive/1791151586.920494",
                "schema": "ArchiveResult",
                "start_ts": "2026-10-04T22:10:19.180742+00:00",
                "status": "succeeded"
            }
        ],
        "wget": [
            {
                "cmd": [
                    "wget",
                    "--no-verbose",
                    "--adjust-extension",
                    "--convert-links",
                    "--force-directories",
                    "--span-hosts",
                    "--no-parent",
                    "-e",
                    "robots=off",
                    "--reject=js,mjs,woff,woff2,ttf,otf,eot,mp4,m4v,mov,webm,ogv,mkv,avi,flv",
                    "--timeout=120",
                    "--restrict-file-names=windows",
                    "--page-requisites",
                    "--user-agent=Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/118.0.0.0 Safari/537.36 ArchiveBox/0.7.4 (+https://github.com/ArchiveBox/ArchiveBox/) wget/GNU Wget 1.21.3",
                    "--compression=auto",
                    "--timestamping",
                    "https://github.com/Niko1221/Strata"
                ],
                "cmd_version": "GNU Wget 1.21.3",
                "end_ts": "2026-10-04T22:10:19.168518+00:00",
                "index_texts": null,
                "output": "github.com/Niko1221/Strata.html",
                "pwd": "/data/archive/1791151586.920494",
                "schema": "ArchiveResult",
                "start_ts": "2026-10-04T22:10:07.748250+00:00",
                "status": "succeeded"
            }
        ]
    },
    "icons": null,
    "is_archived": true,
    "is_static": false,
    "latest": {
        "archive_org": null,
        "dom": null,
        "favicon": "favicon.ico",
        "git": null,
        "media": null,
        "pdf": null,
        "screenshot": null,
        "singlefile": null,
        "title": "GitHub - Niko1221/Strata: 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. \u00b7 GitHub",
        "warc": null,
        "wget": "github.com/Niko1221/Strata.html"
    },
    "link_dir": "/data/archive/1791151586.920494",
    "newest_archive_date": "2026-10-04T22:10:23.719744+00:00",
    "num_failures": 0,
    "num_outputs": 6,
    "oldest_archive_date": "2026-10-04T22:10:06.459749+00:00",
    "path": "/Niko1221/Strata",
    "schema": "Link",
    "scheme": "https",
    "snapshot_id": "cb886f8b-2d21-425a-8ad4-d2fc547bf8bb",
    "sources": [
        "/data/sources/1791151586-import.txt"
    ],
    "tags": null,
    "tags_str": "",
    "timestamp": "1791151586.920494",
    "title": "GitHub - Niko1221/Strata: 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. \u00b7 GitHub",
    "updated": "2026-10-04T22:10:06.458865+00:00",
    "updated_date": "2026-10-04 22:10",
    "url": "https://github.com/Niko1221/Strata"
}