MTS AI releases Cotype Pro 3 and Cotype Light 3: how it trains models for agentic skills

In July 2026, MTS AI released the third generation of its Cotype language models: the flagship Cotype Pro 3 and the lightweight Cotype Light 3. With the release, the company explained how it collects real task-execution traces to train its models in agentic skills, meaning the ability to get through multi-step scenarios without failures.

What MTS AI released

MTS AI has updated its Cotype lineup to the third generation: the flagship Cotype Pro 3 and the lighter Cotype Light 3, which came out slightly ahead of the flagship. The company positions both models as a key component of enterprise AI agents and multi-agent systems, that is, as the "brain" that drives chains of actions rather than simply answering one-off queries.

• Developer: MTS AI
• Flagship model: Cotype Pro 3
• Lightweight version: Cotype Light 3 (released slightly ahead of the flagship)
• Purpose: the core of enterprise AI agents and multi-agent systems
• Training focus: getting through multi-step scenarios without errors

Why the data can't be made up

Traces are recorded trajectories of how a model solves a real task step by step: it calls a tool, reads the result, and picks the next step. MTS AI trains Cotype's agentic skills on exactly this kind of data, and, as the company stresses in the title of its write-up, this data "can't be made up."

Synthetic data for single-turn answers is easy to generate, but realistic multi-step scenarios are not: they depend on genuine edge cases, tool errors, and decision forks that cannot be reproduced artificially. As a result, an agent's quality depends not on model size but on the quality and authenticity of the collected traces. According to MTS AI, without this kind of training a model "stumbles" on long chains of actions.

Where Cotype 3 is used

Cotype Pro 3 and Cotype Light 3 are intended primarily for enterprise scenarios, as an executor and orchestrator inside multi-agent systems. In such an architecture, one model breaks a task into steps, calls the necessary tools, and passes intermediate results to other agents, so a single error at any step is costly.

MTS AI calls reliability in multi-step scenarios a mandatory requirement for its AI agents to work. The lightweight Cotype Light 3 is designed for more economical deployment, while the flagship Pro 3 is meant for the most complex tasks.

"...how we teach our models to carry out multi-step scenarios without hiccups, which is absolutely essential" — from the MTS AI technical blog on

Habr.

What this means

Russian LLMs are increasingly being tuned not for "chatter" but for work in agentic systems, where the reliability of long chains of actions matters. MTS AI's bet on collecting real traces shows that competition is shifting from model size to the quality of training data for agentic skills.

FAQ

What are Cotype Pro 3 and Cotype Light 3?

They are the third generation of MTS AI language models: Cotype Pro 3 is the flagship version, and Cotype Light 3 is the lightweight one. Both are built as a key component of enterprise AI agents and multi-agent systems.

Why does MTS AI collect traces when training Cotype?

Traces are real step-by-step trajectories of solving tasks. MTS AI trains agentic skills on them so that the model gets through multi-step scenarios without failures; according to the company, such data cannot be reliably synthesized.