@@ -5328,6 +5349,62 @@ async function addOR(id, name) {
toast('Added: ' + name);
}
+// ─── Ollama Cloud model fetcher ───
+let ocModels = [];
+async function fetchOCModels() {
+ const el = document.getElementById('oc-model-list');
+ el.textContent = 'Fetching from ollama.com...';
+ const r = await fetch('/api/admin/ollama-cloud/models');
+ const data = await r.json();
+ ocModels = data.models || [];
+ if (data.error) { el.textContent = 'Error: '+data.error; return; }
+ renderOCModels();
+}
+function renderOCModels() {
+ const q = (document.getElementById('oc-search').value || '').toLowerCase();
+ const filtered = q ? ocModels.filter(m => m.name.toLowerCase().includes(q)) : ocModels;
+ const el = document.getElementById('oc-model-list');
+ if (!filtered.length) { el.textContent = 'No models found.'; return; }
+ const existing = new Set((config.cloud_models||[]).map(m=>m.id));
+ el.textContent = '';
+ filtered.forEach(function(m) {
+ const added = existing.has('ollama_cloud::'+m.id);
+ const row = document.createElement('div');
+ row.className = 'model-row';
+ const nameEl = document.createElement('span');
+ nameEl.className = 'name';
+ nameEl.textContent = m.name;
+ const meta = document.createElement('span');
+ meta.className = 'meta';
+ meta.textContent = m.size_gb + 'GB';
+ row.appendChild(nameEl);
+ row.appendChild(meta);
+ if (added) {
+ const btn = document.createElement('button');
+ btn.className = 'btn btn-sm';
+ btn.disabled = true;
+ btn.style.opacity = '0.4';
+ btn.textContent = 'Added';
+ row.appendChild(btn);
+ } else {
+ const btn = document.createElement('button');
+ btn.className = 'btn btn-sm btn-green';
+ btn.textContent = 'Add';
+ btn.onclick = function() { addOC(m.id, m.name); };
+ row.appendChild(btn);
+ }
+ el.appendChild(row);
+ });
+}
+function filterOC() { renderOCModels(); }
+async function addOC(id, name) {
+ config.cloud_models = config.cloud_models || [];
+ config.cloud_models.push({id: 'ollama_cloud::'+id, display_name: 'Ollama: '+name, enabled: true});
+ await saveCloudModels();
+ renderOCModels();
+ toast('Added: ' + name);
+}
+
async function saveTimeouts() {
var g = parseInt(document.getElementById('global-timeout').value) || 300;
config.timeouts = config.timeouts || {};
@@ -6493,6 +6570,26 @@ def query_ollama(model, prompt, timeout):
return resp.json()["response"]
+def query_ollama_cloud(model, prompt, timeout):
+ """Query Ollama Cloud (ollama.com) — same API as local but with bearer auth."""
+ cfg = load_config()
+ prov = cfg["providers"].get("ollama_cloud", {})
+ base = prov.get("base_url", "https://ollama.com")
+ api_key = get_api_key("ollama_cloud")
+ headers = {"Authorization": f"Bearer {api_key}", "Content-Type": "application/json"}
+ prompt_tokens = estimate_tokens(prompt)
+ ctx_limit = get_context_limit(model)
+ num_ctx = min(max(prompt_tokens + 1024, 2048), ctx_limit)
+ if prompt_tokens > ctx_limit - 512:
+ prompt = smart_truncate(prompt, ctx_limit - 512)
+ resp = requests.post(f"{base}/api/chat", headers=headers, json={
+ "model": model, "messages": [{"role": "user", "content": prompt}],
+ "stream": False, "options": {"num_ctx": num_ctx}
+ }, timeout=timeout)
+ resp.raise_for_status()
+ return resp.json()["message"]["content"]
+
+
# ─── MODEL RATE-LIMIT TIMEOUT SYSTEM ─────────────────────────
# Models that get 429'd are auto-disabled until admin re-enables them.
_model_rate_limited = {} # model_id -> {"since": timestamp, "reason": str, "count": int}
@@ -6573,6 +6670,8 @@ def query_model(model_id, prompt):
provider_name, model_name = model_id.split("::", 1)
if provider_name == "anthropic":
return query_anthropic(model_name, prompt, timeout)
+ if provider_name == "ollama_cloud":
+ return query_ollama_cloud(model_name, prompt, timeout)
return query_openai_compatible(model_name, prompt, provider_name, timeout)
return query_ollama(model_id, prompt, timeout)
except requests.exceptions.HTTPError as e:
@@ -7063,6 +7162,12 @@ def admin_test_provider():
r = requests.get(f"{prov.get('base_url', 'http://localhost:11434')}/api/tags", timeout=5)
count = len(r.json().get("models", []))
return jsonify({"ok": True, "message": f"Connected. {count} models available."})
+ elif name == "ollama_cloud":
+ key = data.get("api_key") or get_api_key("ollama_cloud")
+ base = prov.get("base_url", "https://ollama.com")
+ r = requests.get(f"{base}/api/tags", headers={"Authorization": f"Bearer {key}"}, timeout=10)
+ count = len(r.json().get("models", []))
+ return jsonify({"ok": True, "message": f"Connected to Ollama Cloud. {count} models available."})
elif name == "openrouter":
key = data.get("api_key") or get_api_key("openrouter")
r = requests.get(f"{prov.get('base_url', 'https://openrouter.ai/api/v1')}/models",
@@ -7113,6 +7218,36 @@ def admin_openrouter_models():
return jsonify({"models": [], "error": str(e)})
+_oc_models_cache = {"data": None, "ts": 0}
+
+@app.route("/api/admin/ollama-cloud/models")
+@admin_required
+def admin_ollama_cloud_models():
+ import time
+ now = time.time()
+ if _oc_models_cache["data"] and now - _oc_models_cache["ts"] < 300:
+ return jsonify({"models": _oc_models_cache["data"]})
+ cfg = load_config()
+ prov = cfg["providers"].get("ollama_cloud", {})
+ base = prov.get("base_url", "https://ollama.com")
+ key = get_api_key("ollama_cloud")
+ headers = {"Authorization": f"Bearer {key}"} if key else {}
+ try:
+ r = requests.get(f"{base}/api/tags", headers=headers, timeout=15)
+ r.raise_for_status()
+ models = []
+ for m in r.json().get("models", []):
+ name = m.get("name", "")
+ size_gb = round(m.get("size", 0) / 1e9, 1)
+ models.append({"id": name, "name": name, "size_gb": size_gb,
+ "modified": m.get("modified_at", "")[:10]})
+ _oc_models_cache["data"] = models
+ _oc_models_cache["ts"] = now
+ return jsonify({"models": models})
+ except Exception as e:
+ return jsonify({"models": [], "error": str(e)})
+
+
@app.route("/api/admin/ollama-models")
@admin_required
def admin_ollama_models():