OpenRouter and roleplay
OpenRouter for roleplay on a phone: compatible apps, popular models, and free-tier caps
OpenRouter isn’t a roleplay app—it’s an API gateway with a browser playground. To use it on a phone, generate an OpenRouter API key and paste it into a client that supports it, such as SillyTavern, Tavo, or Foreverse (Android and iPhone). OpenRouter says it adds zero markup on inference, charges 5.5% ($0.80 minimum) when buying credits, and limits :free models to 50 requests a day before you top up.
Available for iPhone and iPad in selected regions outside mainland China. Not available on the mainland China App Store. Download help
Does OpenRouter have an app?
As far as we could find, OpenRouter doesn’t ship a mobile app of its own—none appears on its website or inside its documentation. Its web chat page is a side-by-side comparison playground spanning 600+ models, and the “Apps” directory on openrouter.ai simply showcases third-party tools that route API traffic through OpenRouter.
So when people ask for an “OpenRouter roleplay app,” they really want a mobile frontend with an OpenRouter key slot. You sign up on OpenRouter once, copy one API key, and drop it into a client like Foreverse (which ships OpenRouter as a built-in provider). From there, one key reaches paid and :free models alike, and you can switch models without touching the app’s billing.
Which roleplay apps run on OpenRouter?
OpenRouter publishes live token leaderboards under Entertainment → Roleplay. Here is how the “Today” chart looked when we opened it on 2026-11-15, sorted by single-day token volume. Remember: token traffic tells you how busy a platform is, not whether you’ll like using it.
| Rank | App | Tokens that day | Description on OpenRouter |
|---|---|---|---|
| 1 | Hello Minds | 77.1B | Powered by Ethoswarm (hellominds.ai) |
| 2 | Janitor AI | 33.6B | Chatbot platform to create and chat with custom AI characters |
| 3 | Craft | 14.9B | Build and play AI RPGs |
| 4 | SillyTavern | 13.7B | LLM frontend for power users with deep control over characters and prompts |
| 5 | Halluna | 8.31B | halluna.ai |
| 8 | Chub AI | 3.62B | GenAI for everyone |
| 11 | Tavo | 1.27B | tavoai.dev |
Among those entries, SillyTavern and Tavo are dedicated clients you can point at your own key from a phone; see our wider guide to BYOK roleplay apps for more mobile options.
What is the best model for roleplay on OpenRouter?
OpenRouter puts a plain-English disclaimer right on its charts: rankings “measure usage on OpenRouter and reflect adoption, not model quality or benchmark performance.” Keep that caveat front and center. Across OpenRouter’s Roleplay collection (rolling seven-day tokens, November 2026), DeepSeek V4.1 Flash 1017 led at ~525B tokens (23.4% share), followed by DeepSeek V4 Flash (156B), Xiaomi’s MiMo V2.6 Pro (135B), a second DeepSeek V4.1 Flash listing (129B), Z.ai’s GLM 5.3 Flash (125B), and OpenAI’s GPT-6 Luna (110B).
Zoom into individual apps and habits diverge fast. Monthly totals on OpenRouter’s SillyTavern page put Gemini 4 Argon first (48.6B tokens), ahead of DeepSeek V4.1 Flash 1017 (31.5B) and GLM 5.3 Flash (28B). Over on Janitor AI’s page, the top three were DeepSeek V3.2 (144B), Nemotron 3 Ultra (121B), and GLM 5.2 (56.3B). Again—crowd favorites, not a blind writing test.
Price changes the picture too. When we checked on 2026-11-15, OpenRouter listed DeepSeek V4.1 Flash at $0.0507 per million input tokens and $0.2028 per million output tokens (tagged “83% off”), compared with $0.30 / $1.20 peak rates on DeepSeek’s own API. Meanwhile, Google prices Gemini 4 Argon at $2 / $10 per million tokens through November 30, 2026 and $4 / $20 starting December 1, meaning the most-used SillyTavern model doubles in price within weeks. Always check live model cards before settling in.
How do I use OpenRouter in a roleplay app on my phone?
- Generate an OpenRouter API key
Sign up on OpenRouter and create a key on its API keys page. That single key unlocks both
:freeand paid models. - Choose free-only or load prepaid credits
Models ending in
:freework at $0. Paid models draw from prepaid credits; OpenRouter takes a 5.5% fee ($0.80 minimum) at top-up and states it passes upstream inference prices through with zero markup. - Connect OpenRouter inside your phone app
In Foreverse, go to Me → AI Models & Services, pick OpenRouter, paste the key, and tap Test connection. Don’t see a brand-new model in the picker? Add its exact OpenRouter model ID manually.
- Cap your key and verify the first few turns
Put a spending limit on the key inside OpenRouter’s dashboard, send two or three test messages, and cross-check OpenRouter’s activity log against your app’s request records.
Running into 401, 404, or 429 errors? Our BYOK troubleshooting guide walks through each status code, while OpenRouter’s own docs also describe a separate 402 error for credit limits.
Are OpenRouter’s free models any good for roleplay?
OpenRouter hosts $0 :free endpoints and rate limits them. On 2026-11-15, the four busiest free models were NVIDIA Nemotron 3 Ultra, NVIDIA Nemotron 3.5 Super VL, Baidu ERNIE-5.5-Flash-Preview, and NVIDIA Nemotron 3.5 Lightning—all pitched in OpenRouter’s blurbs toward reasoning, coding, agents, and document work; we found no mention of roleplay in those descriptions. We haven’t tested those four on character cards, so we won’t guess how they write.
| Lifetime credits purchased | Requests per minute | Requests per day |
|---|---|---|
| Less than $10 | 20 | 50 |
| $10 or more (the higher cap starts at $9 to absorb top-up fees) | 20 | 1,000 |
Every reply and every swipe counts as a request, so 50 calls a day vanishes in a single afternoon chat. Pointing your model ID at openrouter/free auto-routes across free endpoints, at the cost of never knowing which model will answer next.
When OpenRouter is not the right route
- You stick to one model company. Getting a key straight from DeepSeek, Anthropic, or Google skips OpenRouter’s 5.5% top-up fee.
- You want free models to carry multi-hour daily chats. Without a $10 lifetime top-up, 50 free requests a day runs out fast.
- You expect leaderboards to pick the best writer. OpenRouter explicitly notes its charts track token volume, not prose quality.
- You want an all-in-one roleplay app built by OpenRouter. OpenRouter only provides the API plumbing.
FAQ
Does OpenRouter have a roleplay app?
No—we found no mobile app built by OpenRouter itself. OpenRouter is an API gateway plus a web comparison playground; apps like SillyTavern, Janitor AI, and Tavo show up on its leaderboard because they route user requests through OpenRouter’s API.
What is the best model for roleplay on OpenRouter?
OpenRouter tracks token volume rather than writing quality. In November 2026, DeepSeek V4.1 Flash 1017 led its overall Roleplay collection, whereas SillyTavern users spent the most tokens on Gemini 4 Argon. Audition two or three models on your own card to see which voice fits.
Are there free models on OpenRouter for roleplay?
Yes, any model ID ending in :free costs $0, capped at 20 requests/minute and 50 requests/day until you’ve bought $10 in lifetime credits (which raises the daily cap to 1,000). Most top free models are positioned for coding, reasoning, or agents rather than fiction.
Does OpenRouter charge extra on top of model prices?
According to OpenRouter’s FAQ, inference prices pass through from providers with zero markup, while credit purchases carry a 5.5% fee ($0.80 minimum).
Can I use OpenRouter on my phone?
Yes—just paste an OpenRouter key into a mobile client that supports it. Foreverse includes OpenRouter as a built-in preset, and Tavo also appears on OpenRouter’s Roleplay app leaderboard.
Sources
- OpenRouter: Roleplay app rankings (checked 2026-11-15)
- OpenRouter: Roleplay model collection (checked 2026-11-15)
- OpenRouter: Free models collection (checked 2026-11-15)
- OpenRouter docs: Limits (checked 2026-11-15)
- OpenRouter docs: FAQ (checked 2026-11-15)
- OpenRouter: SillyTavern usage page (checked 2026-11-15)
- OpenRouter: Janitor AI usage page (checked 2026-11-15)
- OpenRouter: Chat playground (checked 2026-11-15)
- DeepSeek API docs: Models & Pricing (checked 2026-11-15)
- Google AI for Developers: Gemini API pricing (checked 2026-11-15)