Ranked by headcount · 25 of 40 on the roster
Builds the Vision Pro spatial computer and visionOS — Apple's bet that head-mounted displays become the next personal computing platform. The M5 Vision Pro starts at $3,499 with 2x on-device AI inference performance.
Releases the open-weight Llama model family including Code Llama, which enables on-premise and self-hosted AI coding tools for enterprises with data-residency requirements.
Creates Claude — the frontier LLM family that powers Claude Code, the terminal-based coding agent that became the breakout developer tool of 2025.
Glean is an enterprise AI platform that combines connected work search with AI assistants and agents across a company's apps and data.
Builds frontier large language models including GPT and Codex that power the majority of AI coding tools via API, plus first-party products like ChatGPT and Codex CLI.
Cohere builds enterprise large language models and AI agents focused on regulated industries and sovereign deployments.
Perplexity is an AI answer engine that delivers cited, conversational answers and is expanding into an agentic browser, Comet.
xAI is Elon Musk's frontier AI lab, building the Grok family of large language models and the Colossus supercomputer.
Zhipu AI is a Tsinghua-incubated Chinese large-model developer behind the GLM family and the Z.ai platform.
01.AI is Kai-Fu Lee's Beijing large-model startup behind the open Yi model family, now focused on commercial AI solutions.
AI21 Labs is an Israeli foundation-model company building enterprise LLMs and AI systems, including the Jamba model family.
Baichuan Intelligence is a Beijing large-model startup founded by ex-Sogou CEO Wang Xiaochuan, building the open Baichuan model series.
MiniMax is a Shanghai-based foundation-model lab behind the ABAB language models and the Hailuo video and audio generators.
European foundation model lab producing the Mistral and Codestral model families with competitive coding performance and a strong focus on sovereignty and on-premise deployment for EU enterprises.
Moonshot AI is a Beijing foundation-model lab behind the Kimi chatbot and the open-weight Kimi K2 model family.
StepFun is a Shanghai multimodal foundation-model lab founded by ex-Microsoft executive Jiang Daxin, behind the Step model family.
Creator of Devin, the first autonomous AI software engineer that can plan, write, test, and deploy complete software projects from natural-language specifications.
Decart builds real-time, interactive world models and the inference stack that runs them.
DeepSeek is a Hangzhou AI lab spun out of the High-Flyer quant fund, known for the low-cost open-weight DeepSeek-V3 and R1 models.
Genspark builds a general-purpose 'Super Agent' that turns natural-language prompts into completed multi-step tasks.
H Company is a Paris-based AI lab building agentic foundation models and autonomous agents for enterprise and consumer automation.
Hark builds multimodal AI models and AI-native hardware devices designed to be a universal interface between humans and machines.
Imbue is an AI research lab building reasoning models and tools that let AI agents reliably code and complete tasks.
Liquid AI builds general-purpose foundation models based on liquid neural networks, an alternative to transformer architectures.
AI lab building foundation models purpose-trained for software engineering, with a model architecture designed specifically for code generation and understanding rather than general-purpose conversation.
