offgrid-ai 0.3.20 → 0.3.21

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "offgrid-ai",
-  "version": "0.3.20",
+  "version": "0.3.21",
   "description": "Privacy-first CLI for running local LLMs — discover, configure, run, benchmark",
   "author": "Eeshan Srivastava (https://eeshans.com)",
   "type": "module",

package/src/backends.mjs

@@ -72,16 +72,18 @@ async function scanOllamaModels() {
     if (!response.ok) return [];
     const body = await response.json();
     if (!Array.isArray(body?.models)) return [];
-    return body.models.map((model) => ({
-      id: model.name,
-      label: ollamaLabel(model.name),
-      aliasSuggestion: model.name,
-      sizeBytes: model.size ?? 0,
-      quant: model.details?.quantization_level,
-      family: model.details?.family,
-      backend: "ollama",
-      source: "ollama",
-    })).sort((a, b) => a.label.localeCompare(b.label));
+    return body.models
+      .filter((model) => isLocalOllamaModel(model))
+      .map((model) => ({
+        id: model.name,
+        label: ollamaLabel(model.name),
+        aliasSuggestion: model.name,
+        sizeBytes: model.size ?? 0,
+        quant: model.details?.quantization_level,
+        family: model.details?.family,
+        backend: "ollama",
+        source: "ollama",
+      })).sort((a, b) => a.label.localeCompare(b.label));
   } catch {
     return [];
   }
@@ -112,6 +114,13 @@ async function scanOmlxModels() {
 
 // ── Labels ──────────────────────────────────────────────────────────────
 
+function isLocalOllamaModel(model) {
+  const name = String(model?.name ?? "");
+  if (/:cloud(?:$|\b)/i.test(name)) return false;
+  if (!Number.isFinite(model?.size) || model.size <= 0) return false;
+  return true;
+}
+
 function ollamaLabel(name) {
   return name.replace(/[-_]/g, " ").replace(/^gemma\b/i, "Gemma").replace(/^qwen/i, "Qwen");
 }

package/src/cli.mjs

@@ -280,7 +280,7 @@ async function printModelCatalog({ profiles, newModels, managedItems }, items =
     const backendItems = managedItems.filter((item) => item.backendId === backendId);
     if (backendItems.length === 0) continue;
     const be = BACKENDS[backendId];
-    console.log("\n" + pc.bold(`Available from ${be.label}`));
+    console.log("\n" + pc.bold(`Local models via ${be.label}`));
     for (const { model } of backendItems.slice(0, 10)) {
       const num = itemNumber((item) => item.type === "managed" && item.backendId === backendId && item.model.id === model.id);
       console.log(managedModelCard(num, model, be));
@@ -307,13 +307,13 @@ function downloadedModelCard(num, model, caps) {
     ["Good for", humanCapabilitySummary(caps)],
     ["Size", formatBytes(model.sizeBytes)],
     ["When selected", "offgrid-ai will recommend safe local settings"],
-  ]), { formatBorder: caps.mtp ? pc.blue : pc.yellow });
+  ]), { formatBorder: pc.yellow });
 }
 
 function managedModelCard(num, model, backend) {
   return renderCard(`${num}. ${model.label}`, renderRows([
-    ["Status", statusText("info", "Available from another app")],
-    ["App", backend.label],
+    ["Status", statusText("info", `Local model via ${backend.label}`)],
+    ["Runs with", backend.label],
     ["Model ID", pc.cyan(model.id)],
     ...(model.quant ? [["Size/type", model.quant]] : []),
   ]), { formatBorder: pc.magenta });

package/src/profile-setup.mjs

@@ -36,14 +36,14 @@ export async function configureLocalProfile(prompt, profile) {
   console.log(pc.dim("You can accept the recommended settings. Bigger conversation memory uses more RAM.\n"));
 
   if (caps.mtp) {
-    console.log(renderSection("Speed boost available", "This model supports fast draft mode. It can make responses feel faster while keeping everything local.\n\nAdvanced: uses llama.cpp MTP on port 8081."));
-    const useMtp = await prompt.yesNo("Use fast draft mode for this model?", true);
+    console.log(renderSection("MTP available", "This model supports multi-token prediction (MTP). offgrid-ai can run it with llama.cpp MTP on port 8081."));
+    const useMtp = await prompt.yesNo("Use MTP for this model?", true);
     configured = useMtp ? applyMtpDefaults(configured) : removeMtpDefaults(configured);
   }
 
   if (caps.qat) {
     console.log("");
-    console.log(renderSection("Optimized model", "This model was trained to work well after compression. No extra runtime settings are needed."));
+    console.log(renderSection("QAT model", "This model is marked as quantization-aware trained (QAT). No extra runtime settings are needed."));
   }
 
   if (caps.thinking) {

package/src/ui.mjs

@@ -83,8 +83,8 @@ export function humanCapabilitySummary(caps = {}) {
   const parts = [];
   if (caps.thinking) parts.push(pc.magenta("Reasoning"));
   if (caps.vision) parts.push(pc.cyan("Vision"));
-  if (caps.mtp) parts.push(pc.blue("Fast draft mode"));
-  if (caps.qat) parts.push(pc.green("Optimized quantization"));
+  if (caps.mtp) parts.push(pc.blue("MTP"));
+  if (caps.qat) parts.push(pc.green("QAT"));
   return parts.length > 0 ? parts.join(" · ") : "General chat";
 }