Fireworks: Ember-1

Benchable lists the Fireworks endpoint fireworks/ember-1-20260923 with the same author description: a specialized reasoning model from Fireworks Research built on Kimi K3, producing shorter reasoning traces using roughly 40% fewer tokens while maintaining comparable quality. The model accepts image and text input and p The Benchable page records a 1M context window, a release date of September 23, 2026, and pricing of $3 per million prompt tokens, $15 per million completion tokens, and $0.30 per million input cache read tokens. The benchmark results table on this page is empty, with no executed evaluations available. The Fireworks en

Fireworks' Ember-1 is described as a specialized reasoning model from Fireworks Research built on Kimi K3, designed to produce shorter reasoning traces using roughly 40% fewer tokens than the base model while maintaining comparable quality, per its listing on LM Market Cap. The model is positioned for coding, knowledge The aggregator records Ember-1 with a 1.0M token context window (with a 943.7K max output), API pricing of $3 per million input tokens and $15 per million output tokens, and a release date of September 24, 2026. Capability, pricing, context, recency, and output capacity each contribute to the composite signal scores (8