Text, images, video and audio
Every AI model in one subscription
GPT, Claude, Gemini, DeepSeek, Grok, Flux, Nano Banana, Kling, Veo, Suno, ElevenLabs and more — in one subscription. Pay with Russian bank cards or SBP (Faster Payments System), no VPN.
- Russian cards and SBP
- No VPN or foreign cards
- Price shown before you run
- OpenAI-compatible API
Every AI model in one place
Text
Audio and music
- GPT-5.6
- GPT-5.6 Luna
- Claude Opus 5.5
- Claude Sonnet 5.5
- Claude Haiku 4.5
- Gemini 3.1 Pro
- Gemini 3.8 Flash
- Grok 4.7
- DeepSeek V4 Pro
- Qwen3.8 Max
- DeepSeek V4.1 Flash
- Qwen3.8 Flash
- OpenAI Embeddings 3 Small
- Nano Banana 2
- Nano Banana Pro
- FLUX.2 [pro]
- Ideogram V4.5
- Seedream 5.0 Pro
- GPT Image 2
- Kling 3.0 Pro
- Veo 3.1
- Veo 3.1 Fast
- Seedance 2.0
- Hailuo 2.3
- Suno V6
- ElevenLabs v4 (озвучка)
- Whisper
- ElevenLabs Scribe v2
Models
The best AI models in one place
Switch between models within a single conversation instead of juggling a dozen subscriptions and accounts.
Chat, code, documents and analysis
- Claude Opus 5.5
Самая мощная модель Anthropic: глубокий анализ, код, длинные документы.
2.53 / 12.66 MT per 1K tokens
- New
- Claude Sonnet 5.5
Баланс ума и скорости от Anthropic: тексты, код, работа с файлами.
1.26 / 6.33 MT per 1K tokens
- New
- Gemini 3.8 Flash
Быстрая мультимодальная модель Google: текст, картинки, аудио, видео.
0.47 / 2.37 MT per 1K tokens
- New
- Grok 4.7
Флагман xAI для кода, агентов и работы со знаниями.
1.26 / 3.8 MT per 1K tokens
- New
Image generation and editing
- Ideogram V4.5
Лучшая работа с текстом и типографикой на картинках.
from 19 MT per image
- New
- Nano Banana 2
Gemini 3.1 Flash Image: генерация и редактирование картинок до 4K.
from 50.66 MT per image
- Image input
- Nano Banana Pro
Gemini 3 Pro Image: максимум качества, точный текст на картинках.
from 95 MT per image
- Image input
- FLUX.2 [pro]
Фотореалистичная генерация от Black Forest Labs.
19 MT per image
Clips from a prompt or a picture
- Kling 3.0 Pro
Кинематографичное видео 1080p до 15 секунд, со звуком и по первому кадру.
from 70.93 MT per second
- Image input
- Video with sound
- Veo 3.1
Видео Google со звуком и диалогами, 720p–4K.
from 126.66 MT per second
- Image input
- Video with sound
- Veo 3.1 Fast
Быстрая и недорогая версия Veo 3.1 со звуком.
from 63.33 MT per second
- Image input
- Video with sound
- Seedance 2.0
Видео ByteDance со звуком: плавное движение, 480p–1080p.
from 51.3 MT per second
- Image input
- Video with sound
Music, voiceover and speech recognition
- ElevenLabs v4 (озвучка)
Самый выразительный синтез речи ElevenLabs, 90+ языков.
50.66 MT per 1,000 characters
- New
- Suno V6
Песни с вокалом или инструментал по описанию или своим стихам (2 варианта за раз).
38 MT per track
- Whisper
Распознавание речи OpenAI Whisper: аудио и видео в текст.
8.55 MT per minute
- ElevenLabs Scribe v2
Точное распознавание речи с разметкой говорящих, 90+ языков.
5.06 MT per minute
This is just part of the catalog — the list of models is updated regularly.
How it works
One balance — every AI model
Three steps from sign-up to a finished result or an API connection.
- Step 1
Mesh tokens — one currency
Everything is charged in tokens: a chat message, an image, a second of video, a minute of music.
- One balance for text, images, video and audio
- The price in tokens is shown before you run
- If a generation fails, your tokens are returned
- Step 2
Subscription or top-up
Pick the way to get tokens that suits you — or combine both.
- Subscription: tokens every month, plus plan discounts and perks
- Top-up: a one-off token pack, no subscription
- Purchased tokens never expire
- Step 3
API for your projects
Plug AI models into your code, bots and services using the familiar OpenAI format.
- Change only the base_url and key — your SDKs stay the same
- A key with a token limit protects you from overspending
- A dedicated balance lets you hand a key to a client or colleague
Pricing
Plans and top-ups
Choose a subscription or top up your balance when you need to. Tokens are shared across all AI models.
For developers
An OpenAI-compatible API
Change the base_url and key — and your code, SDK, Cursor, LangChain or n8n will work with ModelMesh models.
- OpenAI format: chat/completions, streaming, tools and vision
- A token limit for each key or a dedicated balance
- The cost of every request is in the usage.cost_mt field of the response
- Text, images, audio and video through a single API
https://api.modelmesh.shop/v1Request example
from openai import OpenAI
client = OpenAI(
base_url="https://api.modelmesh.shop/v1",
api_key="mm_live_...", # your ModelMesh key
)
response = client.chat.completions.create(
model="gpt-5-mini", # any model from GET /v1/models
messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)FAQ
Frequently asked questions
A quick guide to tokens, payment and the API. Can’t find an answer — message support.
What are Mesh tokens?
A Mesh token (MT) is the service’s internal currency. Everything is charged in tokens: a chat message, an image, a second of video, a minute of music. Mesh tokens are not language-model tokens: a single “currency” works for all AI models at once, and you see the cost of an operation before you run it.
How much do AI requests cost?
The price depends on the model and parameters: for text — on the size of the request and response, for images — per generation, for video — per second, for voiceover — per character. ModelMesh shows the price in tokens before you run. A plan may give a discount on generations, and on some plans “light” and “medium” text models are available without spending tokens within the request-rate limit stated in the plan description. Which group a model belongs to is shown in the model catalog and when you pick a model.
Do tokens expire?
It depends on the source. Bonus tokens (for sign-up, from promo codes) are valid for a limited time — the expiry is stated when they are credited. Subscription tokens roll over to the next month and period by default while the subscription is active; when a subscription ends without renewal, the remaining subscription tokens expire — while it is active, the money for them can be refunded under the refund policy. Tokens bought through a top-up, including the gift tokens of a pack, never expire. They are spent in the order of “what expires first”: bonus tokens first, then subscription tokens, then purchased ones.
How is a subscription different from a top-up?
A subscription credits tokens every month and unlocks perks: generation discounts, queue priority, more API keys. A top-up is a one-off purchase of a token pack with no subscription or commitment. You can combine both: there is one balance, and tokens of different origin simply add up.
How do I pay from Russia?
With a bank card from a Russian bank or via SBP (Faster Payments System). A receipt under Federal Law No. 54-FZ will be sent to your email. No foreign card or VPN is needed.
Can I get a refund?
The refund terms are described in the refund policy — the link is at the bottom of the site. If a generation fails for a technical reason, tokens are automatically returned to your balance. For anything else, message support — we will sort it out.
How do I cancel a subscription?
In your account, in the “Subscription” section. Auto-renewal will be turned off, and plan perks are kept until the end of the paid period. If you want a refund for months that have not started and unused subscription tokens, message support before the end of the period — the calculation is described in the refund policy.
Is the API compatible with OpenAI?
Yes. ModelMesh follows the OpenAI API format: just set the base_url to https://api.modelmesh.shop/v1 and use your ModelMesh key. It works with the official SDKs and tools such as Cursor, LangChain and n8n.
Can I limit spending per API key?
Yes, a key has two modes. “Limit” spends your account’s shared balance, but no more than the set number of tokens — in total, per day or per month. “Dedicated balance” — you transfer the needed number of tokens to the key, and it spends only those; the remainder returns when the key is revoked. This is handy for issuing keys to colleagues and clients.
Does the service work in Russia, and do I need a VPN?
The service works from Russia without a VPN: we send the requests to foreign AI providers ourselves. Payment is by Russian cards and via SBP.
Try every AI model in one account
Passwordless sign-up by email. Bonus tokens for your first requests.