Solar Pro 3
Neural Network
Solar Pro 3 is an Upstage language model based on Mixture-of-Experts architecture for reasoning, text, and code tasks.
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 131 072 tokens
- Max answer length
- 117 964 tokens
- Reasoning effort
- none, minimal, low, medium, high
- Capabilities
- reasoning, function calling
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Solar Pro 3?
Solar Pro 3 is an Upstage language model built on the Mixture-of-Experts architecture: out of 102 billion parameters, about 12 billion are activated per pass, combining high quality with efficient computation. The architecture is tailored for tasks requiring sequential reasoning, precise text processing, and code analysis. In BotHub, the model is available without a VPN or foreign card: you pay in rubles for actual tokens used, and they do not expire. Other neural networks work alongside in the same window, so you can compare answers or switch in a second. A unified OpenAI-compatible API allows you to connect Solar Pro 3 to your service and switch models later without rewriting the integration, while data is protected by AES-GCM encryption. For teams, there is a contract, invoice, and admin panel with limits. Useful when you need to analyze someone else's code and explain its logic, summarize a long document, draft analytics, automate support responses via API, or test a hypothesis across multiple models simultaneously.Frequently asked questions about Solar Pro 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.
It is a text model with a 128,000 token context: it analyzes long documents, writes and refactors code, helps with debugging, and solves reasoning and STEM-logic tasks. Suitable for analytics, technical documentation, and agent scenarios via function calling.
The output limit is up to 117,964 tokens per response, which is a lot even for voluminous materials. You can get an entire technical section, a large code module, or a detailed document analysis without breaking the task into dozens of separate requests.
Yes, the reasoning mode is confirmed: the model builds intermediate steps before the final answer. This significantly helps with math, logic puzzles, code analysis, and multi-step questions where it is important not to guess the answer, but to carefully reach it through a chain of thought.
Yes, the model supports function calling and structured JSON output. You describe the schema and tools, and receive a predictable machine-readable result for parsing. In BotHub, this works via a unified OpenAI-compatible API, so integration does not require a separate SDK.
Images and documents — yes, the model accepts them as input and responds with text: it recognizes image content, analyzes diagrams, tables, and files. Audio and video input is not noted in our data. You get text as output; the model does not generate media.
We do not have specific data on the quality of Russian for this model, so we cannot promise accuracy — please test it with your task. You can switch to another of the 250+ BotHub models in the same window and via the same API, without rewriting the integration.