OpenRouter CEO: Why Chinese Open Models Are Beating the US | Why Enterprises Fear OpenAI & Anthropic

19 insights saved from this video by @business

@business· Business Models
LoRAs and cartridges let users move customizations to new base models without paying the full retraining cost.
@business· Business Models
On OpenRouter, Luna’s price fell 10x in two weeks while usage jumped 13x.
@business· Business Models
The emerging model-lab market may lose roughly half its independent companies through failure or consolidation.
@business· Business Models
July saw 70 model launches, averaging one new model every 10 hours.
@business· Business Models
Token prices fell 90% in roughly 18 months, forcing inference businesses to reconcile cheaper usage with percentage-based revenue.
@business· Business Models
The same model can perform very differently across inference providers as hardware and serving optimizations change.
@business· Business Models
Model aggregators turn discovery into a product by surfacing capable models users would never choose themselves.
@business· Business Models
The strongest architecture pairs one frontier orchestrator with multiple specialized, low-cost sub-agents.
@business· Business Models
Distillation exposes model outputs for inspection, making alignment problems easier to detect during reinforcement-learning rollouts.
@business· Business Models
AI customers keep underestimating inference demand, making unplanned capacity, failover, and uptime the dominant opportunity.
@business· Business Models
A routing layer should re-benchmark providers continuously and redirect traffic when quality, speed, or price improves.
@business· Business Models
Open-weight models are better suited than frontier models for cheap, focused, deterministic tasks.
@business· Business Models
Creativity is not verifiable by a simple number. That makes model diversity strategically valuable.
@business· Business Models
Harnesses are likely to persist because they shape the user experience and let non-model companies own the developer relationship.
@business· Business Models
American AI development remains behind Chinese open-model development in both quality and pace.
@business· Business Models
Model choice ultimately follows trust in outputs, often built through personal tests instead of standardized benchmarks.
@business· Business Models
Chinese open-weight providers have a structural edge: China concentrates funding, policy, and resources behind national champions.
@business· Business Models
Open-weight models created an inference ecosystem beyond hyperscalers, with Fireworks and Together serving models users rarely run there.
@business· Business Models
No single AI model will win the whole market. The opportunity is specializing in useful capabilities.

More from 20VC with Harry Stebbings

See everything from 20VC with Harry Stebbings

Post on X

Keep what you found in OpenRouter CEO: Why Chinese Open Models Are Beating the US |…

A free account saves any of these cards to your Collection, and Korva resurfaces them so you actually remember.