Text models

GPT-5.6 Luna is speed and efficiency. Search and reasoning are controlled independently; reasoning is off by default.

GPT-5.6 Terra is a versatile choice for most tasks. Search and reasoning are controlled independently; reasoning is off by default.

GPT-5.6 Sol is for the most demanding tasks. Search and reasoning are controlled independently; reasoning is off by default.

A fast Google model for everyday requests, image analysis, and tasks where response speed matters.

Anthropic's flagship model for precise writing, long documents, and professional work.

The xAI model for fresh context, web search, and reasoned answers.

Google model for multimodal context, search, and reasoning.

A model for structured analysis and tasks where answer logic matters.

DeepSeek's reasoning model for tasks that benefit from a detailed solution path.

GigaChat is Sber's language model. It is not available in Neiron; GPT-5.6 Terra covers similar tasks.

Amazon Nova Premier is the senior model of the Amazon Nova family for analyzing text, images, and video inside the AWS stack; it returns text.

Claude Opus 4.8 accepts text and images and returns text. Anthropic still lists the model as available, but marks it as legacy.

Claude Opus 5 is Anthropic's advanced model for long-horizon agentic work and coding; its official overview lists up to a 1M-token context and 128K-token output.

Command A is Cohere's model for tool use, agentic scenarios, and multilingual text tasks.

Llama 4 Maverick is Meta's open Mixture-of-Experts model for general text tasks and image understanding, with weights available for study and deployment.

Llama 4 Scout is Meta's compact open model from the Llama 4 family for fast text tasks and image analysis with lighter resource needs.

Mistral Large 3 is a multimodal Mistral AI model that accepts text and images and returns text.

Kimi K2 Thinking is an open reasoning model from Moonshot AI for competition-grade problems, long reasoning chains, and large contexts.

GigaChat 2 Max is Sber's senior GigaChat model for demanding text tasks, document analysis, and Russian-language work. Beyond text, it understands images and audio.

T-pro IT 1.0 provides open Russian-language weights from T-Tech for further fine-tuning and self-hosting. Its model card explicitly says it is not a ready-to-use conversational assistant.

Suno Lyrics by Suno can create or revise song lyrics. Input: Text. Output: Text.

GPT-5.2 by OpenAI can perform multi-step analysis and produce a reasoned answer. Input: Text, Images. Output: Text.

Claude Opus 4.7 by Anthropic can handle coding, code analysis, and agentic development tasks. Input: Text, Images. Output: Text.

Claude Fable 5 by Anthropic can handle coding, code analysis, and agentic development tasks. Input: Text, Images. Output: Text.

Claude Haiku 4.5 by Anthropic can work with text, instructions, and conversational requests. Input: Text, Images. Output: Text.

Claude Opus 4.5 by Anthropic can handle coding, code analysis, and agentic development tasks. Input: Text, Images. Output: Text.

Claude Opus 4.6 by Anthropic can handle coding, code analysis, and agentic development tasks. Input: Text, Images. Output: Text.

Claude Sonnet 4.5 by Anthropic can handle coding, code analysis, and agentic development tasks. Input: Text, Images. Output: Text.

GPT Codex by OpenAI can handle coding, code analysis, and agentic development tasks. Input: Text, Images. Output: Text.

Gemini 2.5 Pro OpenAI-compatible by Google can perform multi-step analysis and produce a reasoned answer. Input: Text, Images, Audio, Video. Output: Text.

Gemini 3 Pro OpenAI-compatible by Google can perform multi-step analysis and produce a reasoned answer. Input: Text, Images, Audio, Video. Output: Text.

Gemini 3.1 Pro OpenAI-compatible by Google can perform multi-step analysis and produce a reasoned answer. Input: Text, Images, Audio, Video. Output: Text.

Gemini 2.5 Flash OpenAI-compatible by Google can work with text, instructions, and conversational requests. Input: Text, Images, Audio, Video. Output: Text.

Gemini 3 Flash OpenAI-compatible by Google can perform multi-step analysis and produce a reasoned answer. Input: Text, Images, Audio, Video. Output: Text.

Gemini 3.5 Flash by Google can handle coding, code analysis, and agentic development tasks. Input: Text, Images, Audio, Video. Output: Text.

Gemini 3.5 Flash OpenAI-compatible by Google can handle coding, code analysis, and agentic development tasks. Input: Text, Images, Audio, Video. Output: Text.

Gemini 3.6 Flash OpenAI-compatible by Google can handle coding, code analysis, and agentic development tasks. Input: Text, Images, Audio, Video. Output: Text.

Gemini 3 Flash v1beta by Google can perform multi-step analysis and produce a reasoned answer. Input: Text, Images, Audio, Video. Output: Text.

Gemini 3.7 Flash by Google can handle coding, code analysis, and agentic development tasks. Input: Text, Images, Audio, Video. Output: Text.

Gemini 3.7 Flash OpenAI-compatible by Google can handle coding, code analysis, and agentic development tasks. Input: Text, Images, Audio, Video. Output: Text.

Grok 4.6 by xAI can handle coding, code analysis, and agentic development tasks. Input: Text. Output: Text.