Fireworks' Ember-1 reasoning model arrives on Vercel's AI Gateway
Vercel has added Ember-1 from Fireworks to its AI Gateway. Ember-1 is a research preview reasoning model built on Kimi K3 for coding and agentic workflows, and Fireworks says it generates approximately 40% fewer tokens than Kimi K3 at comparable quality.
- Ember-1 from Fireworks is now available on Vercel's AI Gateway as a research preview.
- It is a reasoning model built on Kimi K3 for coding and agentic workflows.
- Fireworks reports approximately 40% fewer generated tokens than Kimi K3 at comparable quality.
- It supports a 1M-token context window, text and image input, tool calling and implicit prompt caching.
- The preview has an initial two-week window; use fireworks/ember-1 as the model name.
What Ember-1 offers
According to Vercel, Ember-1 is a research preview reasoning model built on Kimi K3 and aimed at coding and agentic workflows. Fireworks reports approximately 40% fewer generated tokens than Kimi K3 at comparable quality across its evaluations. These are figures reported by Fireworks, not independent measurements.
The model supports a 1M-token context window, text and image input, tool calling, and implicit prompt caching. The Fireworks endpoint also supports Zero Data Retention and No Prompt Training.
Vercel points to the practical benefit for coding agents that make repeated model calls. Shorter reasoning traces can reduce output costs and the amount of context carried into later steps.
Availability and limits
Ember-1 is offered as a research preview with an initial two-week window. The source does not give pricing or say what happens after that period, so developers who want to test it should plan around the short timeframe.
How to get started
Developers can call the model by using the name fireworks/ember-1 across APIs and coding agents. The source lists support for the AI SDK, Claude Code, Codex and the Chat Completions API. Vercel's example uses the streamText function from the ai package with the prompt "Write a retry helper with exponential backoff."
To use Ember-1 in a coding agent, Vercel gives these steps:
- Install the Vercel CLI with npm i -g vercel@latest.
- Run vercel ai-gateway setup.
- Select fireworks/ember-1 in the agent's model configuration.
According to Vercel, the setup command detects installed agents, provisions or reuses an AI Gateway API key, and configures their connection to the gateway.
What it means for developers
AI Gateway offers a unified API for calling models, with built-in usage and cost tracking, budgets for API keys, and routing rules. Ember-1 can be tried in the model playground. For teams running coding agents, the release adds another reasoning model option that is reachable through the same gateway setup, with the token savings claim from Fireworks as the main reason to test it.
Vercel: Ember-1 from Fireworks now available on AI Gateway
