Distribution, hosting and standards

Distribution, hosting and standards

Hugging FaceThe hub for open models. NVIDIA agreed to buy it for $12.9B on 3 Sep.OpenRouterUpdatedOne API for 400+ models. Its usage rankings work as the popularity chart.Together · FireworksThe leading independent hosts for open models.Cloud model platformsUpdatedWhere enterprises buy models and agent infrastructure under their existing cloud contracts.Ollama · LM StudioRun open models on your own machine.MCPThe universal standard for connecting agents to tools and data.Agent SkillsA portable format for packaged know-how, adopted by 30+ tools.A2A · AGENTS.md · ACPUpdatedStandards for how agents talk to each other, read project instructions and plug into editors.BasetenUpdated“AWS for inference”: $1.5B Series F at $13B (closed); $600M run-rate.Prime IntellectReinforcement-learning platform that lets companies train their own agents; $1B valuation.ZenitySecurity and governance for enterprise AI agents; raised $125M in August.ModalNewServerless AI cloud for inference, agent sandboxes and training: past $1B run-rate, $750M at $15.75B (14 Oct).ArenaNewThe crowdsourced model leaderboard, now a commercial evaluation company: $200M at $3.1B (8 Oct).TemporalNewDurable execution, the plumbing that keeps long-running AI agents reliable; $550M at $12.55B (14 Sep), $250M+ annualized revenue.InferactNewThe company behind vLLM, the most-used open inference engine; $150M seed at $800M, now partnering with Google TPU and AMD.

Compiled from public sources by Pro Pluribus; updated 16 Oct 2026. General information, not advice. Company names and logos belong to their owners and are used to identify them, without affiliation or endorsement. Terms