How can we help?

Okran is Ulter's unified AI platform. Instead of choosing one AI lab, you get the world's frontier models — GPT‑5, Gemini 3.1 Pro, DeepSeek‑V3.2 and Grok 4.3 — in a single conversation, plus an agentic coding companion (Okran Code), Agent Mode and memory.

Okran's signature feature is the Okran Council: four models deliberating over your question, with the best‑suited one answering.

1. Go to okran.ai and choose Sign up.
2. Register with your email and a password (minimum 6 characters), then verify your email from your inbox.
3. Sign in — your chats, settings and memory sync to your account.

Forgot your password? Use Reset password on the sign‑in screen and follow the email link.

Type in the input bar at the bottom and press Send (or Enter). Your messages appear on the right; Okran's responses on the left. Your first message becomes the chat's title, so a descriptive opening line keeps your sidebar organized.

Previous chats live in the sidebar, grouped by date (Today, Yesterday, Previous 7 Days, Older). Click any title to reload the full conversation and continue where you left off.

Use the model selector at the top of the chat to pick a specific model, the Okran Council, or Okran's automatic routing, which reads your question and sends it to the model best suited to answer. The responding model is always disclosed — ask "which model answered this?" any time.

Anything you type in the prompt bar on the Ulter homepage is handed straight to Okran: press send and Okran opens with your question already in the input bar, ready to go.

The Council is the first production‑ready AI council: GPT‑5, Gemini 3.1 Pro, DeepSeek‑V3.2 and Grok 4.3 convene over your question, deliberate visibly, and the best‑suited member answers. The deliberation streams alongside the answer, so it costs you no waiting time. Read the full story on the Council page.

1. Open the model selector at the top of the chat.
2. Choose Okran Council — you'll see the four member avatars stacked together.
3. Send your message. The roster appears instantly and the deliberation streams live above the answer.

Default mode is built for speed: the best‑suited member takes your question immediately and the discussion streams in parallel — zero added latency.

Voting mode is the full ceremony: each member casts a visible vote on who should answer, and the tally decides. It's a little slower by design — watching four frontier models argue over your question is the point.

The Council seats members by strength: deep multi‑step reasoning, math, and file analysis go to GPT‑5; code and structured technical work go to DeepSeek‑V3.2; translation and multilingual chats go to Gemini 3.1 Pro; quick everyday questions and current events go to Grok 4.3. The deliberation card shows the reason for every pick — and if a member fails mid‑answer, a fallback chain reseats the question automatically.

Free accounts include 7 Council messages (within the free plan's 30‑message total). Okran Pro removes the cap — see Account & billing.

• GPT‑5 (OpenAI) — deep reasoning, study help, long documents.
• Gemini 3.1 Pro (Google) — multilingual and multimodal strength.
• DeepSeek‑V3.2 — code, algorithms and technical problem‑solving.
• Grok 4.3 (xAI) — instant answers with current, real‑world knowledge.

Model availability can evolve as providers release and deprecate versions; the selector always shows the current lineup.

1. Click the Compare button in the input bar (it highlights when active).
2. Select two or more models from the chips that appear.
3. Send your message — every selected model answers in side‑by‑side cards.
4. Click Choose This on your favorite; that response continues the conversation and the chat switches to that model.

Tip: 2–3 models is the sweet spot; more gets overwhelming. Free accounts include 5 comparisons.

For hard problems, raise the thinking level in the input bar. Higher thinking gives the model more time to reason before answering — in the Council, it also weights selection toward the deep‑reasoning chair (GPT‑5).

1. Click the paperclip in the input bar.
2. Select your image (JPG, PNG, GIF, WebP) or document.
3. A preview appears — add your question and send.

Okran shows its progress ("Analyzing image…", "Reading file…") while it works. Free accounts include 3 uploads.

Okran Code is the agentic coding companion — in the app and on the command line. It reads your project, writes and refactors across files, and explains its changes.

Agent Mode lets Okran browse, research and complete multi‑step tasks (comparisons, planning, preparing bookings) while showing you every step it takes.

Free accounts include 30 messages across all features, and within that total: up to 7 Okran Council messages, 5 model comparisons, and 3 file uploads. Plain single‑model messages have no separate sub‑cap. We state limits plainly — no hidden throttles, no "unlimited" fine print.

Okran Pro unlocks unlimited messages, unlimited Council access, comparisons and uploads. Upgrade from Settings → Plan in the app; current pricing is shown there. Subscriptions renew automatically until cancelled.

1. Open Settings from the bottom of the sidebar.
2. In Account, edit your display name, change your email (verify from your inbox), or set a new password (minimum 6 characters).

Cancel any time from Settings → Plan — one click, effective at the end of the current billing period. Your data and chat history remain intact on the free plan. If anything blocks you, email support@ulter.ai and we'll sort it out.

1. Open Settings → Privacy & Data.
2. Click Export — a JSON file downloads with your profile, every chat session, every message and all timestamps.

Data portability is a basic right, not a premium feature.

Delete individual chats from the sidebar, or clear all history in Settings → Privacy & Data.

To delete your account: Settings → Danger Zone → Delete Account. You'll be asked for your password (case‑sensitive) as a safety check. Deletion removes your chat history and data and cannot be undone.

Okran can remember useful context across conversations — preferences, projects, ongoing work. View, edit or delete individual memories in Settings, or turn memory off entirely. Memory is yours to control.

To generate a response, your message is sent to the model provider selected for that request (OpenAI, Google, DeepSeek or xAI), under their own terms. We never sell your data, and intimate conversations are never used for advertising. Full details in our Privacy Policy.

Every conversation runs through the Graduated Response Framework — safety architecture that detects unhealthy emotional dependency and redirects gently, with warmth instead of rejection. Okran is not a substitute for professional help; if you're in crisis, contact local emergency services or a crisis line immediately.

1. Refresh the page and resend — transient provider hiccups usually clear immediately.
2. Try a different model from the selector, or the Council (it fails over between members automatically).
3. Still stuck? Send us a report from Help & Support in the app.

You've reached a free‑plan cap (30 total messages, or the 7‑Council / 5‑Compare / 3‑upload sub‑caps). Upgrade to Okran Pro in Settings → Plan for unlimited access. Unlike the platforms in our case study, we tell you the exact numbers up front.

• Check the format — JPG, PNG, GIF and WebP are supported.
• Keep files reasonably sized (a few MB); very large files may be rejected.
• Confirm you have free uploads remaining (3 on the free plan).

Make sure the Compare button in the input bar is active (highlighted) and that you've selected at least two models before sending. Then give all cards a moment to finish streaming before choosing a winner.

1. Double‑check your email and password (passwords are case‑sensitive).
2. Use Reset password on the sign‑in page and follow the email link.
3. Check spam for the verification email if your account is new.

In the app, open Help & Support from the sidebar, describe your issue and send — our team replies by email. You can also write to support@ulter.ai directly. Support is real and answered by humans; unanswered tickets are a bug, not a policy.