Claude Sonnet 4
Neural Network
Claude Sonnet 4 is an Anthropic model for coding and complex reasoning, available via the unified BotHub API.
Max answer length
(in tokens)
Context size
(in tokens)
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Claude Sonnet 4?
Claude Sonnet 4 is an Anthropic model that advances the Sonnet line: compared to Sonnet 3.7, it works more accurately with code and reasoning and follows instructions better, achieving 72.7% on the SWE-bench engineering benchmark. Improved controllability means the model adheres more precisely to the specified format and task boundaries, carefully edits existing code, and explains why it proposes a specific solution. In BotHub, Claude Sonnet 4 works without a VPN or foreign card: you pay with a Russian card in rubles only for tokens used, and the remaining balance does not expire. Over 250 models are available in the same window, making it easy to compare answers, and a unified OpenAI-compatible API allows switching models in the multi-model service without rewriting integration; requests are transmitted with AES-GCM encryption. For teams, there is a contract, invoice, EDI, and an admin panel with limits. It is useful when you need to refactor a bloated module, find the cause of a floating bug in logs, write tests and pull request reviews, analyze technical documentation, or go through a rigorous derivation in a math problem with a student.Frequently asked questions about Claude Sonnet 4
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
claude-sonnet-4 works with text, code, and documents: a context of up to 200,000 tokens allows you to keep a large repository or a voluminous contract in one dialogue. Suitable for refactoring and debugging, analyzing long files, STEM tasks, and agent scenarios with tool calling.
In one response, the model outputs up to 64,000 tokens — this is dozens of pages of text or a large code module in its entirety, without splitting into parts. If you need more, continue in the same dialogue: the 200,000-token context will save everything written previously.
Yes, our data shows reasoning and chain-of-thought: before answering, the model breaks down the task step-by-step. This is noticeable in math, logic, finding errors in code, and controversial formulations — the answer turns out more coherent than when generating on the fly.
Yes. Function calling and structured JSON output are confirmed, so the model is easy to integrate into an agent or workflow: it will call your functions and return the result according to the specified schema. In BotHub, this is available via a unified OpenAI-compatible API.
Images and documents — yes: upload a screenshot, diagram, PDF, or table and ask to analyze the content. Audio and video input is not noted in our data, so for such tasks, it is more convenient to choose a specialized model in the same BotHub window.
There is no formal data on languages in the model card, so we will not make promises for it. The easiest way to check is with your own task: the BotHub interface is in Russian, access is without VPN or foreign cards, payment is in rubles, and neighboring models can be compared in the same window.