Text model

GigaChat — Sber's language models with a 128K context

Q: Which GigaChat models are available in GPTunneL?

The current list is always visible in the «Every GigaChat model» block above — it is built straight from the platform catalogue.

Q: How is GigaChat different from YandexGPT?

Both families are strong in Russian. GigaChat has a noticeably larger context in its top tiers, while YandexGPT ties into Yandex services. Both are available in GPTunneL.

Q: Which tier should I pick?

Pro covers most tasks. Take Max for long documents and hard reasoning — it holds 128 thousand tokens — and Lite for a stream of repetitive requests.

Q: Can GigaChat work with documents?

Yes. The top tiers hold around two hundred pages of text in context, so a contract with its annexes or a full policy can be read in one request.

GigaChat models by Sber in GPTunneL: a 128 thousand token context and native-level Russian. One balance shared with ChatGPT, Claude and Gemini.

Try GigaChat Available models

What the GigaChat family is good at

The traits Sber keeps from generation to generation — you can count on them whichever model you pick.

Russian as the primary language

The key difference from global models: this family was not trained on Russian as an afterthought. It is more precise with grammatical cases, business phrasing and regional context.

Long documents in one pass

A 128 thousand token context is around two hundred pages per request — a contract with its annexes or a full internal policy fits without slicing into chunks.

Three tiers for the budget

Lite for a stream of repetitive requests, Pro for everyday work, Max for hard tasks — no overpaying for the simple ones.

No separate contract

In GPTunneL the models are available right after sign-up, from the same chat as everything else — no API onboarding of your own.

Which model to pick

Three tiers in the family — for different tasks and budgets.

	GigaChat 2 LiteThroughput and speed	GigaChat 2 MaxHard tasks	GigaChat 2 ProEveryday work
Best for	Classification, simple extraction, auto-replies	Analysis, long documents, hard reasoning	Everyday text and correspondence
Context	128K tokens	128K tokens	2K tokens
Speed	Fastest	Medium	Fast
Price per token	Lowest	Highest	Above average

Specs come from the platform catalogue. Exact per-model prices are on the Pricing

Every GigaChat model

Release dates follow Sber's official announcements.

March 2025

GigaChat 2 Max GigaChat 2 Pro GigaChat 2 Lite

The current generation, released in three tiers at once: Lite, Pro and Max. Context grew to 128 thousand tokens — around two hundred pages per request, four times the previous limit.

2023

GigaChat

The first generation and Sber's first large Russian-language model.

How billing works

There is no subscription: you top up one balance and spend it on any model on the platform.

Pay per token

You are charged for exactly what the request and the answer used. Not using it costs nothing: no limits, no monthly fee.

One balance for every model

GigaChat, ChatGPT, Claude, image and video generation — all out of the same wallet. No separate subscription per service.

Top up the way you prefer

International cards, Apple Pay, Google Pay or crypto. The minimum top-up is $5.

See pricing

How to start with GigaChat

One account for every model on the platform.

Emailyou@example.comPassword••••••••

Turn this policy into a short staff briefing — what changed and what to do now

policy.docx

GigaChat 2 Max

Policy briefing

Approval is now a two-step process — the head of department first, then finance.

The response window is down from five working days to two, counted from registration rather than submission.

Requests below the minor-spend threshold go through the simplified form and skip the second approval entirely.

The model switches right inside the input — you can change it mid-conversation.

The answer arrives in the chat and the context is kept until the conversation ends.

Try GigaChat in GPTunneL

Signing up takes a minute, and $5 on the balance is enough to see whether the model fits your task.

Try GigaChat

Frequently asked questions

/01Why use GigaChat through GPTunneL rather than directly?

There is no separate contract and no API onboarding — the model is available right after sign-up, from the same chat as ChatGPT, Claude and image generation, off one shared balance.

/02Which GigaChat models are available in GPTunneL?

/03How is GigaChat different from YandexGPT?

/04Which tier should I pick?

/05Can GigaChat work with documents?

GigaChat is Sber's family of large language models. Its defining trait is the Russian language: the family was not trained on Russian as an afterthought, so it is more precise than translated equivalents with grammatical cases, business phrasing and regional context.

The current GigaChat 2 generation shipped in three tiers — Lite, Pro and Max — and holds up to 128 thousand tokens in context, around two hundred pages of text per request. That is enough to read a contract with its annexes or an entire internal policy without slicing it into chunks.

In GPTunneL the GigaChat models are available right after sign-up — no separate contract and no API onboarding, from the same chat as every other model, with one balance and billing for the tokens you actually spend.