DeepSeek V4.1 Flash Third-Party Test: Did Not Reclaim the Top Spot Among Domestic Models, but Costs Only One-Seventh of Kimi and GLM

Beating AI News Flash: Artificial Analysis has completed independent testing of DeepSeek V4.1 Flash, which scored 40 on the composite intelligence index. It has surpassed its own V4 Pro 0813's score of 36, but still trails Kimi K3's 44 and GLM-5.3's 45. According to this third-party leaderboard, DeepSeek has yet to reclaim the top spot among domestic models.

But the price gap is striking. V4.1 Flash costs an average of just $0.27 to complete one Intelligence Index task, while Kimi K3 and GLM-5.3 both cost around $2 — roughly a 7x difference. Output speed reaches 197 tokens/s, far outpacing Kimi K3's roughly 36 tokens/s and GLM-5.3's roughly 58 tokens/s.

Agent capabilities remain its strongest suit. On AutomationBench-AA, V4.1 Flash scored 69%, matching GPT-6 Astra and surpassing GLM-5.3's 62%; on the long-context test AA-LCR, it also scored 84%. The downside is that it is exceptionally verbose — in AA testing, it averaged about 89,000 tokens of output per task, 62% more than V4 Pro, eating into some of its price advantage.