Morph V3 Fast
Neural Network
Morph V3 Fast is a fast code editing model from Morph with 96% accuracy and a speed of ~10,500 tokens per second.
Specifications
- Input
- text, documents
- Output
- text
- Context size
- 81 920 tokens
- Max answer length
- 38 000 tokens
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Morph V3 Fast?
Morph V3 Fast is a code editing model from Morph: you provide an instruction, source code, and an edit snippet, and receive the modified code. The claimed speed is ~10,500 tokens per second with 96% accuracy, making it suitable for rapid code transformations in task streams. Important: requests must follow a strict format—instruction in an <instruction> tag, source code in a <code> tag, and the edit in an <update> tag, otherwise the model won't understand the structure. Via BotHub, this is accessible from Russia: no VPN or foreign cards needed, pay in rubles per token, tokens do not expire. Over 250 models are available in one window, so you can solve complex tasks with another model and use Morph V3 Fast for edits. It features a unified OpenAI-compatible API, AES-GCM encryption, and corporate terms with contracts, invoices, and limits. Typical scenarios include mass mechanical edits in large repositories, fixing linter errors, renaming and updating signatures project-wide, library migrations, and integrating edits into CI and agentic development pipelines.Frequently asked questions about Morph V3 Fast
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.
morph-v3-fast:batch is a text model with document support: it accepts text and files and returns text. A context of up to 81,920 tokens allows processing large materials, and a response of up to 38,000 tokens is suitable for summarization, data extraction, rewriting, and code work.
The model can output up to 38,000 tokens in a single response. The input context reaches up to 81,920 tokens, so a large document along with instructions fits into one request. If a long result is needed, it is reasonable to break the task into parts and assemble the text in fragments.
There is no separate reasoning mode in our model data, so we cannot confirm it. This note should not be read as a refusal: the absence of a record does not mean the capability is missing. The easiest way to check is with your own tasks—for example, multi-step logic or calculations in documents.
Function calling and structured JSON output are not marked for this model in our data, so we cannot guarantee them. This is not a confirmed refusal—just no label. If the response format is critical, run a test request via the BotHub API and check the result.
The model accepts text and documents as input—files can be processed along with the question. Images, audio, and video are not marked in our data, and support for them is not confirmed. If your task involves images or sound, you can switch to a different model type in the same BotHub window.
There are no separate metrics for Russian language quality in our data, so we will not provide an assessment. The 81,920-token context allows uploading large documents and instructions, and it is best to check how accurately the model handles your specific material using a few real examples.