{"slug": "ekorbia-v0-6-v0-7-a-built-in-model-engine-and-a-visual-refresh", "title": "Ekorbia v0.6 & v0.7 — a built-in model engine, and a visual refresh", "summary": "Ekorbia v0.6 ships a bundled inference engine that runs models locally without requiring Ollama or a terminal, while v0.7 delivers a visual refresh with simpler theme names and quieter defaults. The new engine is macOS-only for now, with Linux and Windows continuing to use Ollama or a custom OpenAI-compatible endpoint. Existing Ollama users can keep their setup, and the app adds a curated model catalog with checksum-verified, resumable downloads.", "body_md": "[← All posts](./index.html)\n\n#\nEkorbia *v0.6 & v0.7* — a built-in model\nengine, and a visual refresh\n\nTwo more releases have landed since the voice-enabled\nv0.5. **v0.6** is the big structural one: Ekorbia now\nruns models itself, with an inference engine that ships\n*inside* the app — no Ollama and no terminal\nrequired. **v0.7** builds on that with a warmer,\ncalmer interface and an easy way to pull almost any model\nstraight from Hugging Face.\n\n## v0.6 — Ekorbia runs models itself\n\nUp through v0.5, Ekorbia was a front-end for\n[Ollama](https://ollama.com)\n— you installed Ollama separately, and Ekorbia talked to\nit. v0.6 changes the default: a bundled inference engine\nships with the app, so a fresh install can chat without\ninstalling anything else first.\n\n### A bundled engine, on by default\n\nFor new installs, the bundled engine is the default.\nFirst launch looks at your Mac's memory, recommends a\nmodel sized to fit, and downloads it with one click —\nyou're chatting a couple of minutes later. No separate\ninstall, no `ollama serve`\n\n, no terminal\nanywhere in the flow.\n\nThe engine is a statically-built\n`llama-server`\n\nthat runs entirely on your\nmachine, on a loopback address, and is supervised by\nEkorbia so it starts and stops cleanly with the app. The\nbundled engine is macOS-only in this release; Linux and\nWindows continue to run against Ollama or a custom\nendpoint (below).\n\n### Already on Ollama? Nothing breaks\n\nIf you were already using Ollama, upgrading doesn't rip\nit out from under you. You get a one-time offer to keep\nOllama or switch to the bundled engine, and either way\nOllama stays available under\n**Settings → Backend**. Your existing models, chats,\nprompts, and settings carry over.\n\n### A built-in model catalog\n\nOn the bundled engine, the model manager shows a curated\n**catalog** — Google's official quantization-aware\nGemma 4 builds (with vision) and the\n`nomic-embed-text`\n\nembedding model that folder\nsearch relies on. Downloads happen in-app with a live\nprogress bar, are **checksum-verified** so a corrupted\nfile never looks installed, and are **resumable** —\ncancel keeps what you've already pulled and picks up\nwhere it left off.\n\n### Bring your own backend\n\nPrefer a different runtime? v0.6 adds support for any\n**OpenAI-compatible** server — LM Studio, llama.cpp's\n`llama-server`\n\n, vLLM, and friends. Point\nEkorbia at it under **Settings → Backend** and it\nspeaks that server's API directly. Between the bundled\nengine, Ollama, and a custom endpoint, Ollama is now one\noption among several rather than a requirement.\n\nOne nicety under the hood: switching backends re-checks your document embeddings, since a different engine produces different vectors. Ekorbia flags attachments indexed on the previous backend and offers the usual one-click re-index, so folder search stays accurate across a switch.\n\n### A visual refresh — and some fixes\n\nv0.6 also refreshes the look. The default\n**One Dark** theme moves to a slate palette with\nbrighter accents (the previous look is preserved as a new\n**Warm Dark** theme), with polish across the chrome —\na glowing model-status pill, tinted capability badges,\nsofter rows, and gentler motion. Alongside that: the\nmessage footer now reports the model's real generation\ntime (so a first message that had to load the model no\nlonger shows an inflated number), the app no longer\nflashes between light and dark at startup, and the macOS\nDock icon sits correctly on Apple's icon grid.\n\n## v0.7 — a friendlier, calmer app\n\nWith the engine story settled, v0.7 is a pass over how\nthe app *feels*: less jargon, quieter defaults,\nand more of the content you actually care about up front.\n\n### Less jargon, warmer defaults\n\nThemes now have plain names (Midnight, Warm, Daylight,\nSlate, Dusk, Sand) plus a **Match System** appearance\nthat follows macOS. The status bar and model pickers read\n**Local · Private** instead of raw engine names. And\nthe per-message token and timing numbers — useful, but\nnoisy by default — now live behind a\n**Show technical details** toggle, one click away on\nany message.\n\n### A content-forward sidebar and new-chat screen\n\nChat-history rows are now roomier cards that show the title, a one-line preview of the latest message, and a relative time — so you can find a conversation at a glance. An empty chat opens with a time-of-day greeting and starter cards that drop one of your saved prompts straight into the composer.\n\n### A simpler Settings panel\n\nSettings goes from seven tabs to six. Keyboard shortcuts\nget their own **Hotkeys** tab (with a one-click\n*Reset to defaults*), and the separate Prompts,\nMemory, and Attachments tabs merge into a single\n**Files** tab. Input fields got clearer too — empty\nfields read as faint hint text, and settings that have a\ndefault show it as a \"Defaults to …\" helper instead of a\nreal-looking value.\n\n### Browse Hugging Face for models\n\nThe catalog gained **Qwen3.5** (4B and 9B builds), but\nthe bigger news is that you're no longer limited to the\ncatalog. Paste a repo id (`org/model`\n\n) into\n**Add from a link** and click **Browse**: Ekorbia\nlists the repo's GGUF files with their sizes and\nquantization, flags a sensible default, pairs a vision\nprojector automatically when one's present, and\n**verifies each download against Hugging Face's\npublished checksums** — the same integrity guarantee as\nthe built-in catalog. Adding a model you already have a\nlink to is a single field now, too.\n\n### Polish, and a v0.7.1 follow-up\n\nv0.7 tidied a lot of small things: the model catalog\nleads with the best pick for your Mac and folds the long\nlist behind a filter, the Watches panel groups summaries\nunder each watch, and the \"thinking\" waiting state got a\nlittle more alive. A quick **v0.7.1** follow-up fixed\na handful of rough edges in comparison chats and the\nsidebar — side-by-side columns no longer blank out when a\nreply finishes and now auto-scroll as they answer, list\nrows update the moment you send or receive a message, and\nthe private-chat button is always available, even inside\na Space.\n\n## How to get it\n\nThe\n[releases page](https://github.com/ekorbia/ekorbia-desktop/releases/latest)\nhas the latest bundles for macOS, Windows, and Linux.\nExisting users upgrade in place — your chats, prompts,\nand settings carry over, and if you're on Ollama you'll\nget a one-time choice to keep it or switch to the bundled\nengine.\n\nThe bundled engine is macOS-only for now; on Windows and Linux, Ekorbia runs against Ollama or any OpenAI-compatible endpoint, and everything else in v0.6 and v0.7 works across all three platforms.", "url": "https://wpnews.pro/news/ekorbia-v0-6-v0-7-a-built-in-model-engine-and-a-visual-refresh", "canonical_source": "https://ekorbia.com/blog/2026-07-25-ekorbia-v0-6-and-v0-7.html", "published_at": "2026-07-25 00:00:00+00:00", "updated_at": "2026-07-25 23:37:44.203860+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-products", "ai-tools", "ai-infrastructure", "developer-tools"], "entities": ["Ekorbia", "Ollama", "Hugging Face", "Google", "Gemma 4", "LM Studio", "llama.cpp", "vLLM"], "alternates": {"html": "https://wpnews.pro/news/ekorbia-v0-6-v0-7-a-built-in-model-engine-and-a-visual-refresh", "markdown": "https://wpnews.pro/news/ekorbia-v0-6-v0-7-a-built-in-model-engine-and-a-visual-refresh.md", "text": "https://wpnews.pro/news/ekorbia-v0-6-v0-7-a-built-in-model-engine-and-a-visual-refresh.txt", "jsonld": "https://wpnews.pro/news/ekorbia-v0-6-v0-7-a-built-in-model-engine-and-a-visual-refresh.jsonld"}}