Zhipu AI's GLM-5.2 model has matched the performance of Anthropic's Mythos on cybersecurity benchmarks while costing only a quarter as much, according to the company. The results highlight a growing trend in the AI industry: performance parity is no longer enough — price is becoming a key differentiator.
Benchmark parity
The two models were evaluated on a set of cybersecurity benchmarks. GLM-5.2 scored on par with Mythos, the data shows. The benchmarks cover a range of common security tasks, though Zhipu AI did not specify which ones. What's clear is that the Chinese lab's model achieved the same level of accuracy and capability as Anthropic's offering.
Cost advantage
At one quarter the cost, GLM-5.2 presents a stark pricing difference. For enterprises and government agencies that need to run large-scale security analyses, the savings could be substantial. Running GLM-5.2 costs about 25% of what it takes to run Mythos for the same tasks. That gap raises questions about how Anthropic and other AI developers will respond. Pricing has become a critical factor as more organizations look to deploy AI for cybersecurity.
The matchup between GLM-5.2 and Mythos is not just about one benchmark. It signals that Zhipu AI is closing the gap with leading Western AI labs. The company has been investing heavily in model development and efficiency. If it can maintain this performance at lower cost, it could capture a significant share of the cybersecurity AI market.
Anthropic has not commented on the comparison. The company's Mythos model is designed with safety constraints that may add to its computational cost. Whether those constraints justify the higher price is a question customers will now have to weigh.
Zhipu AI has not announced a commercial release for GLM-5.2. The model's availability and pricing for customers remain unclear. But the benchmark results are likely to accelerate adoption of cost-efficient AI in security operations. Other labs will be watching closely — and may need to adjust their own pricing strategies.
