# Fireworks API Models, Pricing and Capabilities Canonical HTML page: [Fireworks API Models, Pricing and Capabilities]() Markdown alternative: [Fireworks API Models, Pricing and Capabilities as Markdown]() Fireworks AI is an inference platform serving open-weight models with an emphasis on throughput, fine-tuning, and function calling. Access 1 model from Fireworks through the Impossibl API, including Ember-1. Compare pricing per million tokens, context windows, and capabilities between different Fireworks models. Select two or three models to [compare specifications and API costs](). Model types and billing units are not interchangeable. ## At a glance - models: 1 - modalities: takes text and image in, gives text out - latest release: Ember-1 (Sep 23, 2026) ## Models ### [Ember-1]() Ember-1 is Fireworks Research's post-trained version of Kimi K3, tuned to reason in shorter traces at comparable quality, which Fireworks reports as roughly 40% fewer output tokens. It keeps Kimi K3's long context, image input and tool use, and can reason before answering or answer directly. Its published context window is 1,048,576 tokens. The advertised input rate is $3.00 per million input tokens. The advertised output rate is $15.00 per million output tokens. - API identifier: `fireworks/ember-1` - Creator: Fireworks; model type: Text generation. - Context window: 1,048,576 tokens; tool calling: Not published. - Inputs: text, image; outputs: text. - Base API pricing (USD): $3.00 per million input tokens; $15.00 per million output tokens. - [Complete model facts, cache rates, context tiers, and FAQs as Markdown]() ## Frequently asked questions ### Which Fireworks models are available through Impossibl? The directory lists 1 model from Fireworks available through the Impossibl API, including Ember-1. Open each model page for its API identifier, capabilities, and current pricing. ### How much do Fireworks models cost? Fireworks model prices vary by model and usage. All rates below are published base API rates in USD. For token-priced models, published input rates start at $3.00 per million input tokens. Published output rates start at $15.00 per million output tokens. Input and output minima may refer to different models. Individual model pages include published cache prices and billing units. Applicable fees and workspace-specific pricing can affect the amount paid. ### What is the cheapest Fireworks model? By published input rate, the cheapest Fireworks text generation model on Impossibl is Ember-1 (fireworks/ember-1) at $3.00 per million input tokens and $15.00 per million output tokens. Cheapest by input rate is not always cheapest for a workload; output-heavy or cached requests can favor another model. ### Does Impossibl add a markup to Fireworks pricing? No. Usage is charged at the provider's list price with 0% usage markup. Prepaid credit purchases include a 5% platform fee, and there is no subscription. The rates on this page are Fireworks's published base rates, and usage comes off your balance when the request finishes. ### What context windows do Fireworks models support? Among Fireworks models with a published limit on Impossibl, the largest context window is 1,048,576 tokens, listed for Ember-1. Context limits vary by model and are separate from maximum output limits. ### What inputs and capabilities do Fireworks models support? Available model types include text generation. Published input support includes image and text; support varies by model. Tool-calling support is unspecified for 1 model. ### How do I use Fireworks models through an API? Use the exact model identifier from the model page with the Impossibl API. For example, select fireworks/ember-1 with /v1/chat/completions. Endpoint compatibility depends on the selected model. The API quickstart explains how to create an API key and send a request. ## Sources and related links Prices, capabilities, limits, and availability describe the public Impossibl API offering. An unpublished fact is unknown; a published zero price is zero. Compare matching billing units and check context tiers before estimating costs. - [Public Impossibl model catalog]() - [Full model directory as JSON]() - [Model directory as Markdown]() - [API quickstart]() - [Model documentation]()