Blog · LE PEAK
LE PEAK: a free AI chat with 23 models in one thread
Chat with DeepSeek, Gemma, gpt-oss, Nemotron and more for free, twelve models without an account. How LE PEAK works, its daily quotas, attachments and what it keeps.
LE PEAK is KTwo’s chat: 23 models from five providers in one conversation. Twelve of them are free with no account at all. A free account unlocks the other eleven, each with its own daily allowance, and you can switch models in the middle of a thread to compare answers or hand a problem to a different model.
What’s in it
The models come in six groups. The free ones need nothing; the rest need a free account and are metered per user, per model, per UTC day.
| Group | Models | Account | Daily limit, per model |
|---|---|---|---|
| NVIDIA NIM | Step 5 Preview, DeepSeek V4.1 Flash, GLM 5.3 Flash, Muse Glimmer 30B, Laguna XS 2.1, Nemotron 3.5 Super VL, Nemotron 3.5 Lightning, Nemotron 3 Ultra | None | None (20 messages a minute) |
| OpenRouter Free | gpt-oss 120B Free, Gemma 4 26B A4B, MiniMax M3.1 Flash, North Mini Code | None | None (20 messages a minute) |
| OpenRouter Premium | GPT Luna, Claude Haiku | Free account | 10 requests or 10,000 tokens |
| Xiaomi MiMo | MiMo V2.6 Pro, MiMo V2.6 Flash, MiMo V2.5 Pro, MiMo V2.5 | Free account | 10 requests or 10,000 tokens |
| Ollama Cloud | Beam, gpt-oss 120B | Free account | 40,000 tokens |
| TypeSafe Jev | Jev 1.14.0, Jev Preview, Jev Latest | Free account | 1,000,000 input tokens; output is free |
The line-up changes as providers add and retire models; the model menu in the chat is always current. KTwo’s own models, Crust 1 Beta and Mantle 0.5, are not in LE PEAK: every reply comes from a third-party model.
How to use it
- Open chat.ktwo.dev/peak. No sign-up is needed for the free models.
- Pick a model. Press ⌘K (Ctrl K on Windows and Linux) to open the model selector, laid out like a spec sheet, with each model’s quota.
- Switch whenever you like. Change models mid-thread and the conversation carries on with the new one.
- Read the telemetry. Under each reply: the model, latency, tokens in and out, and how much of your quota is left.
- Sign in for more. Locked models offer Sign in to unlock. Accounts are free.
You can also start from the home page: type a message, pick a model and press Open in LE PEAK. The message arrives in the chat’s message box, ready to send. A link can only fill the box; it can never send a message or spend a quota by itself.
Picking a model
Some starting points:
- Images, video and PDFs: seven of the free models read images, among them DeepSeek V4.1 Flash, Gemma 4 26B A4B and Nemotron 3.5 Super VL. Attach a photo, a screenshot, a video or a scanned PDF and ask about it; see Attach PDFs, images, video and audio.
- Audio: the Xiaomi MiMo models listen to audio clips, with a free account.
- Code: North Mini Code, free on OpenRouter, is a model made for programming.
- Big open models: gpt-oss 120B, free on OpenRouter and also on Ollama Cloud with an account, and Nemotron 3 Ultra, a mixture-of-experts model with 550B parameters, 55B of them active per token. What those numbers mean is explained in Mixture of experts, explained.
- Decisions, not prose: TypeSafe’s Jev models answer with structured decisions, such as a choice or a score, instead of free text.
- Second opinions: ask one model, then switch and ask another the same question in the same thread.
Quotas: per user, per model, per day
The free models have no daily cap, only a guard of 20 messages a minute. The others each have their own daily allowance, counted separately for every model, so using up one never affects another. Allowances reset at 00:00 UTC. When a model runs out, the chat counts down to its reset, and every other model stays available.
How that metering holds up when requests arrive in parallel is the subject of a separate post: Per user, per model, per UTC day.
Privacy: what LE PEAK keeps
- Your messages and files go to the provider of the model you pick, to produce the reply, under that provider’s terms. KTwo doesn’t store them. Your browser prepares attachments first: images are re-encoded, which drops their location and camera data.
- Your chat history is saved in your browser’s local storage, and only there: your latest 100 chats. Attachments aren’t kept, only their names. Delete a chat from the list, or clear the site’s data in your browser to remove them all.
- With an account, KTwo keeps your account ID, your email, the name and username you set (if any), a sign-in session and your usage counts per model for each day you used it (requests and tokens). Passwords are handled by Supabase; KTwo never sees them.
- Rate limits count requests against a salted hash of your IP address, never the address itself.
- No analytics and no trackers.
The details are on the privacy page. Don’t paste secrets into any chat, this one included.
Model names in LE PEAK, and their makers’ logos, are trademarks of their respective owners. KTwo AI Labs doesn’t own them and isn’t affiliated with or endorsed by those companies.
Frequently asked questions
Is LE PEAK free?
Yes. Twelve models are free with no account, limited only to 20 messages a minute. Eleven more are free with an account, each with its own daily allowance.
Do I need an account?
Not for the twelve free models. A free account unlocks GPT Luna, Claude Haiku and the Xiaomi MiMo, Ollama Cloud and TypeSafe Jev models.
Does KTwo store my chats?
No. Messages go to the model’s provider to produce a reply, and your chat history is kept only in your browser.
What happens when I hit a daily limit?
That model is unavailable until 00:00 UTC, when its allowance resets. Every other model stays available, so you can switch and carry on in the same thread.
Can I use KTwo Crust 1 Beta in LE PEAK?
No. LE PEAK serves third-party models only. Crust 1 Beta is in private beta; you can request access.
Keep reading
-
LE PEAK
Attach PDFs, images, video and audio to a free AI chat
LE PEAK now reads your files: ask free AI models about PDFs, photos, screenshots, videos, code and audio. Which models read what, the limits, and what stays private. -
Engineering
Per user, per model, per UTC day: metering a free AI chat
How LE PEAK enforces daily AI quotas with one atomic SQL statement, reserves room for streamed replies, and rate-limits guests without storing IP addresses. -
Research
Mixture of experts, explained: why 120B can run like 5B
How mixture-of-experts AI models work, what names like 30B-A3B and 550B-A55B mean, and the trade-off between total and active parameters in memory, speed and cost.