Marketing and sales
Copy, product descriptions, creatives and client replies: drafts in minutes instead of hours.
Every AI model
in one window
Write texts, parse documents, answer clients — the models take the routine, you check the result. Inside: 1,308 models — text, images, audio; pay only for what you use — no mandatory subscription, no limits.
Start free
No card · ≈$1 of credit for tests right away
1,308 models
llmall
Draft a reply to the client: delivery is 3 days late, keep it polite.
Done: an apology, the new delivery date and an offer to help choose.
deepseek-v4.1-flashdeepseek-v4-flashglm-5.3minimax-m3
Auto-combo picks the cheapest model available
Ask anything…
- Save up to 5,000 ₽ a year
- Pay by SBP or a Russian card
- No five-hour limits
- No VPN
- One API key for the whole catalogue
01 · Who
Who this is for
Product and analytics
Parsing interviews, calls and documents, summaries and hypotheses — in the same window as your work tasks.
Development
Code generation and review, debugging and documentation: one API key and OpenAI-compatible endpoints for every catalog model.
GPU and subscription owners
Your capacity and unused limits become income: the platform brings clients and handles payouts.
02 · included with your account
Ready prompts by task
The prompt library collects tried formulations for typical tasks. No need to write a prompt from scratch: pick one, paste your task — the draft is on screen.
Marketing and sales
- Draft an email to a client
- Product description
- Reply to a review
- Social media post
- Ad copy
Product and analytics
- Interview summary
- Call breakdown
- Feedback digest
- Hypotheses to test
- Interview questions
Development
- Code review
- Explain an error
- Function documentation
- Test cases
- Regular expression
Research and market
- Competitor overview
- Market analysis
- Solution comparison
- Research findings
- Survey questions
Not a blank page, but a ready start
03
Fast inference
Models with the fast suffix answer at real-time dialog and agent speeds. Below: measurements on llmall itself — tokens per second at several concurrent requests.
| model | c=5 | c=6 |
|---|---|---|
| qwen-3.8-27b-fast | 968 | 1182 |
| gpt-oss-120b-fast | 800 | 77 |
Measured on the llmall Gateway: tokens/sec at the given concurrency (c). Real speed depends on the provider and load. · Speed at which a dialog never stalls: you are not waiting for the answer and get more done.
04 · abliterated
Uncensored models
The catalog has abliterated models — some of the refusals and limits imposed during training are lifted. They work through the same API and are billed like ordinary models. Access is granted on request.
More freedom
in scenarios where ordinary models refuse to answer
Same API and billing
as the other models
Available in the common catalog
and billed like ordinary models
Responsibility Responsibility for use is yours: make sure the result meets your rules and applicable law.
05 · for users and providers
Features
Save on top-ups
Top up directly by SBP — no currency conversions or card fees. Zero cost for virtual-card issuance and annual maintenance — that is ≈ 4,000–5,000 ₽ a year you keep.
≈ 4,000–5,000 ₽ a year stays with you
No limits
No 5-hour, weekly or monthly subscription limits. Use the models as much as you need — pay only for tokens actually spent.
No waiting on a limit — you get the answer right now
Auto-combos to save
Ready auto-combos and your own model lists: a request always goes to the cheapest available model in the combo — no manual switching. Build your own or take ready ones — save on token cost without losing quality.
Save on token cost with no loss of quality
Become a provider
Have your own inference capacity? Connect it to llmall, set your own price for model calls — and earn on every request. The platform brings clients and handles payouts.
Earn on every request
Earn during idle time
Do your GPUs sit idle at night and off-hours? Give them to the platform as a provider — while you rest, llmall works for you.
While you rest, llmall works for you
Your subscription starts earning
Case: you have an ollama.com or opencode.ai subscription and the limits sit idle. Become an llmall provider: connect your endpoints and keys, set your price for model calls. llmall users use your models, the platform pays out — you monetize what you already paid for.
You monetize what you already paid for
One provider instead of many
One balance, one API key, one account — instead of a dozen subscriptions, cards and providers. All payout history and per-model stats in one place.
One balance instead of a dozen subscriptions: no separate payments nor tracking token limits in every account
Live statistics
Tokens/sec, uptime, working hours and errors per model. The provider rating is built on real metrics, not promises.
Pick a provider by facts: speed, uptime and errors are visible before you pay
06 · DATA MASKING · SUBSCRIPTION
Personal data never reaches the model
The client sends a request as-is. The service finds names, phone numbers, emails, passports, insurance and tax ids, bank cards, replaces them with markers, and only then forwards the request upstream. The answer comes back and the markers are restored: the user sees their own data, the provider does not.
| Field | What the client sent | What the provider received |
|---|---|---|
| Full name | Иванов Иван Иванович | [FIO_1] |
| Phone | +7 916 000-00-00 | [PHONE_RU_1] |
| client@example.ru | [EMAIL_1] | |
| Passport | 4510 000000 | [PASSPORT_RU_1] |
| Tax id | 770000000000 | [INN_1] |
| Card | 4111 1111 1111 1111 | [CARD_1] |
| Insurance id | 000-000-000 00 | [SNILS_1] |
| IP address | 192.168.1.10 | [IP_1] |
| IBAN | DE89 3704 0044 0532 0130 00 | [IBAN_1] |
| MAC address | 00:1A:2B:3C:4D:5E | [MAC_1] |
| Ethereum wallet | 0x52908400098527886E0F7030069857D2E4169EE7 | [ETH_1] |
| Date of birth | 12.03.1990 | [BIRTHDATE_1] |
| Address | г. Москва, ул. Ленина, д. 5 | [ADDRESS_1] |
The example is fictional. The marker-to-value mapping lives in process memory only and expires within an hour — it is never written to disk.
07 · In roubles, pay as you go
What it costs
Full tables by family —
| Model | Input, ₽ / 1M | Output, ₽ / 1M |
|---|---|---|
| deepseek-v4.1-flash · auto-set | from 14,7 ₽ | from 58,8 ₽ |
| deepseek-v4-flash · auto-set | from 9,36 ₽ | from 22,3 ₽ |
| glm-5.3 · auto-set | from 111,48 ₽ | from 350,36 ₽ |
| minimax-m3 · auto-set | from 29,4 ₽ | from 117,6 ₽ |
| gemini-3.8-flash · auto-set | from 91,88 ₽ | from 459,39 ₽ |
| claude-opus-5.5 · auto-set | from 490,02 ₽ | from 2 450,1 ₽ |
| gpt-6-luna · auto-set | from 9,8 ₽ | from 49 ₽ |
in the dashboard after sign-in
08 · Roubles · VPN · API · subscriptions
Comparing ways to access AI models
The choice is usually not between models but between ways to reach them: a foreign API aggregator, a subscription to one chat service, or a platform that accepts roubles. Below are the verifiable differences; we do not publish others' prices — they change and live in their own catalogues.
| Criterion | llmall | Foreign API aggregator | Chat-service subscription |
|---|---|---|---|
| What you pay with | Roubles: Russian card or SBP | Per the service's rules (usually USD) | Subscription in foreign currency |
| VPN needed | No: portal and API are reachable directly | Depends on your link and the service | Usually yes |
| How the price works | Pay per use | Pay per use | Fixed monthly fee and limits |
| API for apps | Yes, OpenAI-compatible | Yes | Usually no, web UI only |
| What one key covers | 1,308 models of all kinds plus auto-sets | Their catalogue — see their site | One model |
| Spend ceiling | Per-key monthly and daily limits | Per the service's rules | The subscription sets the limits |
Our price is final and includes the platform commission — we do not promise to be the cheapest. Our case: how you pay, no VPN, and one key instead of ten accounts.
09 · How it works
How it works
Create an account
Sign-up in a minute.
Get an API key
One key and OpenAI/Anthropic-compatible endpoints for every catalog model.
Call the models
Pick models in the catalog, watch your balance, statistics and payouts in real time.
10 · OpenAI-compatible API — works with the SDK you already use
Connect in a minute
curl https://llmall.store/v1/chat/completions \
-H "Authorization: Bearer $LLMALL_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"auto-combo/minimax-m3.1-flash",
"messages":[{"role":"user","content":"Привет!"}]}'Point your client at llmall.store/v1 — nothing else changes.
11 · account and combos
What it looks like
The account gathers models, balance, call history and payouts in one window. An auto-combo is a model list from which the platform always picks the cheapest available: if a model is busy, the request goes to the next by price. Build and tune your own combos for your tasks.
The platform counts and routes — you do the work
12 · for GPU and subscription owners
Earn on llmall
three ways to turn idle capacity and subscriptions into income. The platform brings clients and handles payouts.
Your GPU capacity
income from every request
Use subscriptions (ollama.com, opencode.ai and others)
monetize at your price
Night and weekend idle time
the platform works for you
13 · FAQ