Relace Apply 3
Neural Network
Relace Apply 3 is a Relace model for applying code edits, with a Relace Apply 3 API for integration into editors and pipelines.
Max answer length
(in tokens)
Context size
(in tokens)
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Relace Apply 3?
Relace Apply 3 is a specialized model from Relace with a narrow task: take a proposed edit and carefully embed it into the source file without affecting the rest of the code. It is not a conversationalist, but an execution unit: one model devises the change, and Relace Apply 3 transfers it to the files, saving you from manual fragment stitching and insertion errors. In BotHub, it works from Russia without a VPN or foreign card; payment is made with Russian cards in rubles on a pay-as-you-go basis, for tokens, without a subscription. Access is via a unified OpenAI-compatible API, so you can connect the model to an editor or CI without separate integration, and switch to any of over 250 models in one window with a couple of lines of configuration. Data is encrypted and not saved, and teams have access to contracts, invoices, electronic document management (EDO), and an admin panel with limits. Useful if you are building your own code agent, automating repository refactoring, embedding auto-apply edits into code reviews, or building a pipeline where one model generates a patch and another applies it.Frequently asked questions about Relace Apply 3
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.
relace-apply-3 is a text model from Relace, tailored for working with code: it quickly applies ready-made edits to files and assembles the final version in its entirety. Context is 256,000 tokens, it accepts text, documents, and images as input, so the model is suitable for code editors, edit automation, and parsing large files.
Maximum 128,000 tokens per response — a very large volume, enough to return a large file in its entirety after edits, rather than stitching the result from fragments. For short responses, the limit does not interfere: in BotHub, you pay as you go for the tokens used.
A separate reasoning mode is not noted in our data, so we will not promise it. The model is focused on fast and accurate application of edits, not on detailed reasoning. If you need reasoning chains, models with such a mode are available nearby in BotHub, and switching takes a couple of seconds.
Support for function calling and structured JSON output is not noted in our data — this is not a refusal, just no confirmation. Test it in your scenario via the unified OpenAI-compatible BotHub API: if the response format does not suit you, you can replace the model in the request without rewriting the integration.
The model accepts images and documents, this is confirmed by our data, so an interface screenshot or a specification file can be sent along with the code. Audio and video input is not noted; for such tasks, there are separate models in the same window in BotHub.
We have no data on query and response languages, so we will not provide an assessment. It is easiest to check for yourself: send your typical query and compare the result with another model in the same window. Access from Russia without VPN and foreign card, payment with Russian cards.