BriefsNewsPreview
Fireworks announced new AI inference and training products at Forge
Fireworks launched new tools at Forge to help teams build with open-weight models.
Summary
Fireworks announced several new products at its Forge conference, including Ultra-fast inference powered by NVIDIA Blackwell GPUs and Background Mode for lower-cost runtimes. The company also introduced training capabilities like No Eval and Specialized Model Inference, alongside enterprise offerings such as Fireworks Sovereign.
What changed
- Fireworks introduced Ultra-fast inference in public preview, delivering speeds up to 1,500 tokens per second.
- Fireworks launched Background Mode as a lower-cost runtime option at half price for workloads where cost matters.
- Fireworks released No Eval in private preview to turn production traces into task-specific evaluations.
- Fireworks introduced Fireworks Sovereign in private preview to run inference on customer-managed cloud, neocloud, or data centers.
Availability
Preview“It is now available in public preview, delivering speeds up to 1,500 tokens per second as demonstrated on DeepSeek V4.1 Flash .”
Pricing
| Plan | Price, as stated | Checked |
|---|---|---|
| New Pro plans | $8 per month“New Pro plans start at $8 per month, and Teams plans start at $10 per seat per month.” source | |
| Teams plans | $10 per seat per month“New Pro plans start at $8 per month, and Teams plans start at $10 per seat per month.” source |
Prices as stated by the company; we don’t convert or estimate.
Founder takeaway
Test the new inference modes and training tools if you build multi-step agents or want to run open models inside your own infrastructure.
Source
Researched and drafted with AI from the official sources above, then checked automatically before publishing. Spotted a mistake? Tell us.
Related
More briefs
OpenAI shares AI research for physical engineering and hardware design
OpenAI explains how its models help engineers design, simulate, and build physical products across mechanical and electrical fields.
OpenAI1 min read
Krea releases the Krea 3 AI image model for fast editing
The Krea 3 foundation model brings fast, cost-efficient image-to-image editing with support for up to eight reference images.
Krea1 min read
Vercel launches Security Review in public beta with AI vulnerability scans
Vercel released Security Review in public beta to scan code pushes for vulnerabilities using AI-generated severity ratings and agent fixes.
Vercel1 min read
One email, every Monday
Briefs are published here through the day. Monday Heist is the email: one useful AI workflow, the files to use it, and the catch. The first issue is being prepared; subscribers get it first. Free, and you can leave in one click.
One email. Every Monday.