← All companies

Inference / Company profile

Fireworks AI

What Fireworks AI does

Fireworks AI is a B2B inference and model-optimization platform focused on helping teams “own” their AI by deploying, fine-tuning, and composing open(-weight) models in production. Its platform combines an inference layer it describes as optimized for low-latency and high-throughput serving with training and adaptation workflows (including fine-tuning and RL rollouts) so customers can move from experimentation to continuously improving models. Fireworks positions its approach around specialized intelligence—turning general-purpose models into customer-specific behavior—rather than routing workloads purely through commodity model APIs.

News

Company record

Jul 26, 2026 · Official · Fireworks AIFireworks Nexus: Drop-in Open Frontier Intelligence for Teams with Budgets

Fireworks announced Fireworks Nexus, describing it as a drop-in layer for engineering organizations to connect existing tooling to managed open-model inference with enterprise controls and cost observability.

Jul 15, 2026 · Official · Fireworks AIFireworks Secures $1.5 Billion in Series D Funding and Announces $1B ARR

Fireworks announced a $1.505B Series D at a $17.5B valuation, stating it had surpassed $1B in annualized revenue run rate and cited per-day token and specialization figures.

Apr 06, 2026 · Official · Fireworks AIOwn Your AI: Fireworks Training Preview

Fireworks announced Fireworks Training in preview, describing Training Agent, Managed Training, and a Training API as connected surfaces for training and deploying models on shared infrastructure.

Mar 08, 2026 · Official · Fireworks AIFireworks On Microsoft Foundry (Public Preview)

Fireworks announced a public preview of Fireworks on Microsoft Foundry, describing best-in-class open-model inference available through Azure/Foundry.

Mar 08, 2026 · Official · Fireworks AIFireworks Acquires Hathora to Accelerate Global Compute Orchestration

Fireworks announced its acquisition of Hathora, describing Hathora’s latency- and routing-focused global orchestration technology and how Fireworks planned to integrate it.

Oct 28, 2025 · Official · Fireworks AIFireworks AI Raises $250M Series C to Power the Future of Enterprise AI

Fireworks announced a $250M Series C at a $4B valuation, co-led by Lightspeed Venture Partners, Index Ventures, and Evantic, with continued support from Sequoia and strategic participation from several tech/infrastructure firms.

May 29, 2025 · Official · Fireworks AIFireworks DevDay 2025 Wrapped

Fireworks summarized its first Fireworks DevDay 2025, including announcements around optimizers and hardware support highlighted in the event wrap-up.

Jul 11, 2024 · Official · Fireworks AIFireworks Raises $52M Series B to Lead Industry Shift to Compound AI Systems

Fireworks announced a $52M Series B led by Sequoia, citing compound AI systems and an optimized inference platform as core themes.

Show 3 earlier updates
Source map · 4 recurring channels · 12 references

Funding

Latest disclosed valuation$17.5B
Tracked capital$1.8B5 sourced rounds

What it builds

Fireworks Inference Cloud (Inference)

Fireworks provides a hosted inference platform for deploying open-source and specialized models, positioning it as a fast serving stack for production AI workloads.

source ↗
Fireworks Training

Fireworks offers an end-to-end training platform (in preview) spanning multiple training surfaces, including managed training and a training API, aimed at fine-tuning and deploying frontier models at scale.

source ↗
Distributed Inference Engine (Fireworks platform component)

Fireworks describes a disaggregated inference engine architecture optimized for long, multi-turn workloads and multimedia inference (e.g., audio transcription and document/image understanding).

source ↗
Fireworks Developer Toolkit

Fireworks provides a developer toolkit/SDK intended to help developers build and integrate AI applications with models across modalities (text, vision, audio, speech, and video) and support adaptation workflows.

source ↗
Fireworks Nexus

Fireworks Nexus is positioned as a drop-in layer for engineering organizations to connect existing developer tooling to managed open-model inference with enterprise controls and cost observability.

source ↗
Fireworks Training (Training preview surfaces: Training Agent, Managed Training, Training API)

Fireworks Training is described as offering three training surfaces—Training Agent, Managed Training, and Training API—built to run on the same underlying infrastructure that already serves production inference.

source ↗

Key metrics

Milestones & partnerships