Labelbox said on October 14, 2026, that it raised more than $100 million in Series E financing at a $1.25 billion pre-money valuation, in a round led by Infinitum Partners that combines equity with a credit facility.
Wipro Ventures, Andreessen Horowitz (a16z), B Capital, Tau Ventures, and Snowpoint Ventures also participated, among others. Chief executive and co-founder Manu Sharma detailed the terms in a company blog post, and the company said the financing will support continued research and engineering across its two core products, Horizon and Recursion Managed Agents, as well as expansion of the teams bringing those products to customers. Sharma wrote that access to compute will be central to the company’s next chapter, noting that Infinitum’s John Yetimoglu was an early investor in Groq and is now involved in building large-scale compute infrastructure at GroqCloud.
Company Position and History
Labelbox stated that it reached profitability in 2026 while growing more than fivefold year over year. The company reports more than 400 customers, including over 90% of leading U.S. AI labs, and says its base spans frontier AI labs, hyperscalers, startups, and Fortune 500 companies.
Sharma, Brian, and Dan founded Labelbox in 2018. In a founder’s note, Sharma wrote that he came to the United States to learn to design and build aircraft and rockets, working on computational fluid dynamics and neural networks, and that his later roles at DroneDeploy and Planet Labs showed him how much of the opportunity in machine learning sat in data.
The post describes the company’s history in three chapters: training-data software for people building AI; reinforcement-learning environments built after what Sharma calls the ChatGPT moment, when the most capable models began improving by practicing real work; and agents, which the company built for itself first before opening the system to customers.
Sharma also wrote that Labelbox builds each layer of its platform itself: an agent harness designed for ultra-long-horizon agents, sandbox infrastructure that runs millions of virtual machines, its own reinforcement-learning post-training infrastructure, and an expert network of more than 4 million people.
Horizon Reinforcement-Learning Environments
According to the company, Horizon’s WorldSim system builds simulated companies, with code, dashboards, spreadsheets, decks, and documents filled with realistic business data, in which a model attempts a difficult multi-step task end to end while every step is recorded as a trajectory. Deterministic verifiers score each trajectory the same way on every run, and when two trajectories are both correct, an expert’s preference teaches the model what Sharma calls taste. The company describes operating at a scale of thousands of worlds and millions of graded trajectories, tuned to a lab’s pass@k targets across coding, computer use, research, multimodal, and voice work.
The Horizon product page lists domains that include autonomous AI research, agent coding and software engineering, multimodal knowledge work, voice with agentic tool use, computer use, and cybersecurity. Sharma wrote that Labelbox now works with 90% of the leading frontier AI labs and hyperscalers.
Recursion Managed Agents
The funding announcement states that Labelbox opened Recursion Managed Agents to enterprise customers earlier in October 2026. Under the product’s design, a Managed Agent receives a goal, a trigger such as a schedule, an alert, a Slack message, a webhook, or an API call, and a rubric defining what good work looks like. A separate grader agent scores every run against the rubric line by line and points to evidence for each call; when the work misses, the agent revises it.
Agents can operate in three modes: multi-agent, in which a coordinator chains specialists that each carry their own model, instructions, tools, and permissions; team, in which copies of an agent split work on a shared board and check each other’s results; and auto, the default, in which the agent decides for itself when a job needs specialists or a team.
The company says agents work across more than 100 business applications and any MCP server, with every tool scoped as read only, changes data, or destructive and credentials held in a vault so the model never sees them. Sessions run in isolated sandboxes, and Labelbox says the platform runs with zero data retention at every LLM provider. Its listed compliance program covers SOC 2 Type II, GDPR, CCPA, NIST 800-171, and PCI DSS. Pricing is set per agent session hour, with compute and model spend combined in one estimate, and existing Labelbox enterprise customers receive a one-time $100 credit.
Describing the company’s internal rollout in a product launch post dated October 8, 2026, Labelbox wrote, “The surprise wasn’t that agents could do the work. It was how quickly they got better at it.” The company reports running 51 agents across 11 teams, spanning software engineering, reliability, security, machine-learning research, evals, product, finance, sales, strategy, and procurement.
In one internal example, Labelbox said its pull-request review agent ran 39 graded reviews on a single agent version and model, and that across the 33 reviews that passed the grader, the median cost fell from $13.15 for the first 11 reviews to $8.90 for the last 11, 32% less per review, while the agent’s memory was updated 15 times and time per review held at about 27 minutes. Under the product’s memory mechanics, once five new session reports exist, and at most once an hour, a curator agent reads them and updates the agent’s memory; customers can browse what an agent has learned and delete any lesson that no longer holds.
