Odaily reported that Nansen CEO Alex Svanevik has argued that open-weights models may become a stronger competitive force against Anthropic and OpenAI. His point centers on the practical trade-off between model capability and cost, rather than on whether a model sits at the absolute frontier of artificial intelligence performance.
Lower-cost models that are “good enough”
Svanevik said not every task requires a frontier model at a “150 IQ” level. In many use cases, a model with roughly “115 IQ” capability, while costing about 90% less, is already “completely enough.” The comparison frames open-weights models as a cost-efficiency challenge to providers whose business expectations are tied to the most capable frontier systems.
Under this view, the competitive pressure does not come only from matching the most advanced models feature by feature. It comes from the possibility that large numbers of users and tasks can be served by lower-cost models that meet their actual needs. If that happens, the assumption that the highest profits will come from the most advanced frontier models faces a more direct test.
Regulation adds pressure to the business model
Svanevik also linked the issue to government restrictions on frontier AI systems. He said the traditional logic in the AI industry has been that profits would come from the most advanced frontier models, but that logic may be challenged, especially when governments restrict those models or block access to them. In such a setting, revenue expectations for companies built around high-end model access could be affected.
He further questioned whether the commercial path of relying on premium models for profit still holds when regulators begin limiting frontier model capabilities or deployment. According to Svanevik, that question is becoming a core issue the industry needs to reassess as open-weights models gain relevance through lower cost and sufficient performance in many scenarios.