Ember 1
Neural Network
Ember 1 is a reasoning model by Fireworks Research based on Kimi K3 that solves tasks using shorter reasoning chains.
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 1 048 576 tokens
- Max answer length
- 943 718 tokens
- Reasoning effort
- none, low, high, max
- Capabilities
- reasoning, function calling
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Ember 1?
Ember 1 is a specialized reasoning model by Fireworks Research built on Kimi K3. Its concept is to make every token count: reasoning chains are shorter, using about 40% fewer tokens for thinking. For multi-step tasks, this means a more compact solution process while maintaining the analytical approach. In BotHub, the model is accessible without a VPN or foreign card: you pay with a Russian card in rubles for tokens used, and they do not expire. Over 250 models are available in one window, making it easy to switch and compare answers. A unified OpenAI-compatible API allows you to connect Ember 1 to your services and switch models without rewriting the integration. Data is transmitted with AES-GCM encryption and not stored. Companies have access to contracts, invoices, electronic document management, and an admin panel with limits. Useful for solving multi-step tasks, reviewing logic, checking calculations or hypotheses, compiling analytical reports from disparate inputs, or integrating a reasoning model into pipelines where token efficiency is critical.Frequently asked questions about Ember 1
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 reasoning before answering, which accepts not only text but also images and documents as input. It is suitable for coding and debugging, analyzing long materials, analytics, STEM tasks, and function calling scenarios.
According to our data, the output limit is 943,718 tokens per response. This is enough for a large code module, a detailed document analysis, or a long structured report that doesn't need to be manually assembled from multiple separate requests.
Yes, the reasoning mode is confirmed by our data: the model breaks down the task step-by-step before providing a result. This significantly helps in mathematics, logic, code analysis, and questions where the solution process is as important as the final formulation.
Yes, the model supports function calling and structured JSON output. It is easy to integrate into agents and pipelines via the unified OpenAI-compatible BotHub API, allowing you to switch to another model if needed without rewriting the integration.
Our data lists images and documents as input formats: the model analyzes diagrams, screenshots, charts, and text files. There are no notes regarding audio or video, so we cannot promise their processing—please test it with your task in the BotHub interface.
We do not publish data on this model's language capabilities, so we cannot make exact promises—it is fastest to check with your own queries. BotHub itself is fully in Russian: access from Russia without a VPN or foreign card, payment with Russian cards.