all the models — AI benchmark observatory
← Models

Ember-1

Fireworks AI · proprietary · ember-1

Ember-1 is a Fireworks Research specialized model built on Moonshot Kimi K3 that shortens reasoning traces (~40% fewer tokens) while matching K3-max quality on Fireworks' evaluations. Research Preview on Fireworks Serverless (model path accounts/fireworks/models/ember-1). MoE with 2.78T parameters, 1040k context, function calling. List pricing $3.00 input / $0.30 cached input / $15.00 output per 1M tokens (same as public Kimi K3 API pricing cited in the launch blog). Self-reported benches from the Fireworks blog: Terminal-Bench 2.1 82.0%, SWE-bench Verified 92.2%, SWE-Interact 20.0% (no catalog id — noted only), DeepSWE 1.1 75.2%, tau2-bench Airline 66%. Model page verified live (Ready) as of catalog add.

SWE-Bench VeTerminal-BenDeepSWE 1.1

Benchmark scores

Pricing

  • Fireworks$3.00 / $15.00

Input / output per 1M tokens

AA metrics

No Artificial Analysis link yet.

Arena Elo

  • No Arena snapshot linked.