Head of Procurement

Fireworks unveils Ember-1 AI model that cuts tokens by half. Fireworks Research has introduced Ember-1, a specialized artificial intelligence model built on the Kimi K3 architecture that reduces token consumption by approximately 40 percent while maintaining answer quality. This release addresses the high costs associated with long reasoning traces in reasoning models, which can consume over 90 percent of tokens and become exponentially expensive in multi-turn agentic workloads. Ember-1 learns to eliminate unnecessary computation while preserving critical self-reflection, enabling efficient reasoning across diverse tasks. The development process involved over 50 training experiments and 200 evaluations, resulting in novel algorithms for token efficiency. Training encompassed software engineering, mathematics, instruction following, tool use, and extended interactions. Fireworks utilized its Serverless Training infrastructure to accelerate research and deployment. Performance assessments confirm Ember-1's superior cost-quality trade-off. On the Specialized Intelligence Index, Ember-1 set a new Pareto frontier for cost versus performance on Doximity's Bedside Bench, outperforming models including GPT-5.6 Sol, GPT-6 Astra, and Claude Opus 5 on cost-per-task metrics. In software engineering evaluations such as SWE-bench Verified, Ember-1 matched Kimi K3's maximum reasoning accuracy at a fraction of the cost, strictly dominating lower-effort K3 configurations. Token reductions ranged from 35 to 50 percent across benchmarks, with cost savings reaching 68 dollars per task on SWE-bench Verified. Live validation supports these results. Customer A/B tests on production coding workloads demonstrated a 35 percent reduction in token usage per task, with Ember-1 sustaining or improving task completion and success rates. One customer has moved Ember-1 to live production with plans to replace base models. Internal testing by Fireworks developers also confirmed seamless operation with no quality degradation alongside significant token savings. Ember-1 is available now as a Research Preview on Fireworks Serverless. The launch initiates a series of specialized models focused on efficiency. Fireworks is also releasing training support, allowing enterprises to build customized, token-efficient models tailored to specific workloads using proprietary data, reinforcing the trend toward specialized intelligence in the open model ecosystem. This news is intelligently aggregated by AI to deliver industry updates efficiently. It does not constitute opinions or advice.

Fireworks AI releases Fireworks Nexus: A drop-in routing and cost-control layer that moves routine coding work to open-weight models. July 28, 2026 Fireworks AI has released Fireworks Nexus, an AI management and routing platform aimed at engineering organizations. It connects the coding tools developers already use to a managed layer of open-weight models. The problem it targets is well documented. Forbes reported that Uber exhausted its entire 2026 AI budget in four months. Claude Code had reached roughly 5,000 engineers after a December rollout. Fireworks cites the same report, noting agentic adoption climbed from about a third of engineers to more than four-fifths in two months. Fireworks frames the underlying issue as a mismatch, not an overspend. Most organizations run routine work at frontier prices. Operational complexity has made switching to open-weight models unattractive for platform teams. Fireworks Nexus. Nexus is composed of three components: 1. Enterprise controls and cost observability: Teams set budgets at the team or company level. They track ROI across models and tools, and enforce policy from one place. Requests run on the Fireworks production inference platform, with US-hosted endpoints, zero data retention, and coverage across 20 global data centers. 2. Workflow continuity: FireConnect is a one-line install that maps harness model slots to Fireworks models. It is released under Apache 2.0 and can be installed from the Fireworks Dashboard in a single command. Claude Code, Codex and OpenCode keep working unchanged. FireConnect runs on Fireworks Serverless APIs that are Anthropic- and OpenAI-compatible, so most tools connect with a base URL and a model ID. 3. Intelligent traffic management: A custom trained model scores each request's difficulty. Routine requests go to a cost-effective open-weight model served by Fireworks. Difficult requests pass through to your existing provider on your own key, which Fireworks states is never stored server-side. The research team reports this typically delivers a 3-5x cost reduction. One important note for anyone planning a rollout. The router is a research preview. It currently routes between Claude Opus 5 and GLM-5.2, so the pass-through path requires an Anthropic key. An all-open configuration routes between Kimi K3 and GLM-5.2 instead. Performance. Fireworks research team has been testing Nexus with dev teams including Notion and Doximity. Preliminary results show a one-third reduction in cost per merged pull request. The research team also reports a blended token rate roughly a quarter of the closed model labs. These are vendor figures from a preview program, and should be read that way. The more useful evidence comes from two independent evaluations Fireworks cites. Faros AI ran 211 real engineering tasks drawn from 12 repositories across seven model-and-harness routes. Claude Code on GLM-5.2 scored 0.568 on a model-based rubric judge. Claude Code on Opus 4.8 scored 0.521. The cost gap was larger than the quality gap: $0.92 per task against $1.76. Cache share was 89.7% and 99.7% respectively, so caching does not explain the result. Faros stresses that the cohort is company-specific and not a universal leaderboard. Arize, in a joint study with Fireworks, evaluated 10 models on 40 Terminal-Bench tasks at six trials each. That is 2,400 runs and $626 of API spend, graded by each task's own test suite. The headline metric is cost per successful task, which counts every failed and retried attempt. Two findings from that study bear directly on the Nexus design. On easy tasks the frontier premium buys nothing: Kimi K2.6 passed 73% where GPT-5.5 passed 69%. On hard tasks only the top tier competes, with GPT-5.5 at 51% and Kimi K3 at 32%. Arize's evaluation harness is open source, so teams can rerun it on their own workloads. Routing simulated over those runs beat every single-model strategy. A deliberate escalation ladder reached $0.525 per successful task while reliably solving 32.3 of 40 tasks. GPT-5.5 alone cost $0.636 and reliably solved 25 of 40. Naive escalation through all ten models cost $1.319, worse than any single model tested. Ladder design is not optional. How teams actually turn it on. There are three practical paths, and they have different blast radii. The FireConnect path is the lowest friction. The installer requires the Claude Code CLI already on PATH, prompts once for a Fireworks API key, registers a plugin marketplace, and writes Claude Code settings with a timestamped backup. It exposes /fireconnect:on, /fireconnect:off, /fireconnect:setup, /fireconnect:models and /fireconnect:set-models. The second path skips the plugin entirely. Set ANTHROPIC_BASE_URL and a Fireworks key, or use the OpenAI-compatible client with a changed base URL and model ID. Arize noted that adding a model to its harness was one config line. That interchangeability is what makes an escalation ladder buildable. The third path is the router itself, which sits in front of an existing frontier contract. Key takeaways. * Independent runs from Faros AI (211 tasks) and Arize (2,400 runs) support routing by difficulty, not by brand. * Fireworks Nexus is a drop-in AI management and routing layer for engineering orgs, announced July 26, 2026. * Three parts ship: enterprise cost controls, FireConnect for workflow continuity, and a difficulty-aware router. * FireConnect is Apache 2.0, installs in one line, and keeps Claude Code, Codex and OpenCode unchanged. * Fireworks quotes 3-5x cost reduction and a 33% drop in cost per merged PR in preliminary customer testing.

Why Prysm Capital, LLC invested in Fireworks: powering the production era of Enterprise AI. July 21, 2026 Nobody at a board meeting asks 'should we use AI?' anymore. Now they ask why the pilot that wowed everyone in March still isn't in production in October, and why the serving bill tripled. Most of the industry has spent the past three years obsessed with building AI models. But a model creates no value until it runs. Every AI application that graduates from demo to production becomes an inference workload, and those workloads are compounding fast. The bottleneck has shifted from model creation to the complex, six-month slog required to reliably serve traffic at a cost acceptable to the CFO. AI spend will represent $1.3 trillion by 2029, growing ~32% year-over-year from 2025-2029, driven largely by agentic AI adoption. As of 2025, 88% of enterprises had reported using AI in at least one business function, up from 55% two years prior. And the open-source model ecosystem, GLM, Moonshot, Qwen, DeepSeek, and others, now trails the closed frontier by as little as a few months, giving enterprises real optionality for the first time. However, actually running these models at scale in production remains the single hardest problem in enterprise AI, largely because meeting the business's demands for latency and reliability requires navigating a complex landscape of constrained GPU compute and enormous serving complexity. It is a hurdle that most teams simply are not equipped to clear on their own, especially as they often lack the rare ML systems expertise needed to solve these infrastructure challenges internally. Fireworks: The Inference Engine for Enterprise AI Prysm Capital, LLC tracked Fireworks for more than a year before its portfolio company Replit, introduced Prysm Capital, LLC to the team. In its first meeting, Lin made it clear they are building frontier training and inference on one platform. Fireworks has built a leading inference platform for enterprises serving and fine-tuning open-weight models in production. On the inference side, Fireworks provides significant cost and performance advantages, delivering roughly 10x better price-per-token economics and up to 5x faster speeds than closed-source alternatives on coding and agentic workloads. Importantly, its capabilities extend beyond inference to custom fine-tuning and reinforcement learning, enabling enterprises to leverage their proprietary data, workflows, context, and more, to better build, serve, and scale models tailored to their specific business needs. Its platform abstracts the complexity of more than 100,000 deployment configuration combinations into one full-stack inference platform spanning post-training, fine-tuning, reinforcement learning, and model serving, providing the tools enterprises need to scale customized models with greater speed and with differentiated economics. Fireworks serves leading enterprise and AI-native customers, including Cursor, Perplexity, Cognition, and Genspark. Annualized revenue run-rate surpassed $1 billion, up from $280 million at its last funding round. Fireworks now processes more than 40 trillion tokens per day, having nearly tripled the daily volume of tokens served on its platform in the same period. That's why Prysm Capital, LLC is proud to participate in Fireworks' $1.5 billion Series D, led by Atreides Management, Index Ventures, and TCV, along with participation from existing investors including Evantic, Lightspeed Venture Partners, and NVIDIA. Its Thesis: Inference and beyond Prysm Capital, LLC always ask: how is this going to be deployed in the real world? Models will keep leapfrogging each other, but Prysm Capital, LLC is prioritizing the essential plumbing that turns AI into a practical business reality. Prysm Capital, LLC started mapping the inference layer eighteen months ago, when Prysm Capital, LLC noticed its portfolio companies' biggest line item was oftentimes serving costs. A coding assistant lives or dies on completion latency. A real-time answer engine can't wait on a slow model. An agent running thousands of steps unravels when the cost-per-token becomes unsustainable. And the pattern is extending beyond AI-natives. Specialized AI is becoming the enterprise default. More companies are moving beyond one-size-fits-all AI and building their own customized models. Fireworks is helping lead that shift. Enterprises are using Fireworks to fine-tune open models on their private data and unique context, creating specialized intelligence for application-specific models that match or outperform generic frontier alternatives on their workloads, while running faster and at a fraction of the cost. 95% of tokens served through Fireworks today are from models that have been specialized. This is the ground on which Fireworks keeps winning enterprise deployments: performance, reliability, and economics that hold up at scale. The Team and the Path Ahead Lin Qiao and her six co-founders have spent their careers on exactly this problem. Qiao and several co-founders were key leaders and engineers behind PyTorch, the open-source framework underpinning much of modern AI. The broader team built and led AI / ML platforms and large-scale production infrastructure across Meta and Google. Few teams have built and operated production ML systems at that scale, and fewer still have turned that experience into a platform others can use. That pedigree shows up in the product. Inference is a crowded, fast-moving market, but Fireworks competes on more than just inference - its advantages compound: custom kernel optimizations that create performance advantages, GPU utilization gains that grow with volume, and a full-stack platform that goes well beyond reselling compute. This platform breadth positions Fireworks as a core layer in an enterprise's AI stack, providing the infrastructure to power tailored solutions at scale across the business. As AI moves from experimentation into production, Fireworks offers exactly what that shift demands: customized open models served with frontier performance, reliability, and economics. Excited to see what comes next as the company invests this new funding into scaling its platform, expanding global infrastructure, deepening its partnerships, and helping more enterprises build specialized intelligence. Disclaimer: For informational purposes only. Not an offer of any securities or of the investment advisory services of Prysm Capital, L.P. References in this piece are provided solely for illustrative purposes to highlight aspects of Prysm Capital's investment approach and do not purport to represent all investments made by Prysm Capital. Past performance is not indicative of future results. May 4, 2026