GLM-5.3-Flash Slays DeepSeek-V4-Flash: Higher Performance Benchmark at Nearly One-Third the Cost

GLM0%
Token-57.6%

Dynamic Insight: Beating AI News Flash: Following the recent open-sourcing of the GLM-5.3 Flash by Intellecta, it directly competed with the DeepSeek-V4-Flash. Both models focus on affordability, coding, and agency, but based on the current publicly available benchmark tests, Intellecta's model is superior.

On Terminal-Bench 2.1, the GLM-5.3 Flash scored 84.3 points, while the DeepSeek-V4-Flash scored 82.7 points; DeepSWE scored 63.4 against 54.4, leading by 9 points. The former natively supports image and video inputs.

The pricing is also very competitive. After a limited-time 50% discount, the GLM-5.3 Flash is priced at $0.075 for regular inputs and $0.25 for outputs per million tokens. Even at its cheapest point, DeepSeek V4 Flash costs $0.22 and $0.66. GLM's input is nearly 3 times cheaper, and output is 2.6 times cheaper.

After the two-week promotion ends, GLM will revert to $0.15 for inputs and $0.5 for outputs, still below DeepSeek's lowest price. Currently, DeepSeek is only cheaper in terms of cache hit pricing.

ソース：BlockBeats

免責事項：現在のコンテンツは第三者の視点に基づくもの、または第三者の視点からAIが直接翻訳したものです。CoinExはコンテンツの信頼性、正確性、独創性を保証するものではなく、CoinExからの投資アドバイスを構成するものではありません。暗号資産の価格変動は急激に変動します。潜在的なリスクにご注意ください。