🤖Mercury 2.5 Drops Below Average on AI Index
Mercury 2.5: Fast, cheap, but not the smartest
TL;DR
Mercury 2.5, released in September 2026, scores 12 on the Artificial Analysis Intelligence Index, below average. Despite this, it's priced competitively and offers fast performance at $0.25 per 1M input tokens.
Mercury 2.5, a proprietary model released in September 2026, scores 12 on the Artificial Analysis Intelligence Index, placing it below average in intelligence. Despite this, Mercury 2.5 offers a compelling value proposition with its fast performance and competitive pricing. For developers looking for a cost-effective solution with quick turnaround, Mercury 2.5 might be a viable option, especially for tasks that don't require top-tier AI smarts. The model's context window of 260k tokens and verbosity of 35M output tokens make it suitable for extensive text processing tasks. However, the lower intelligence score means it's not ideal for complex cognitive tasks or industries requiring high-level analytical capabilities.
Key Points
Mercury 2.5 released in September 2026, scores 12 on the Artificial Analysis Intelligence Index.
Priced at $0.25 per 1M input tokens, Mercury 2.5 offers competitive pricing.
Output speed of 780.8 tokens per second makes Mercury 2.5 one of the fastest models.
Context window of 260k tokens and verbosity of 35M output tokens support extensive text processing.
Cache discount of 90% reduces costs for frequent users.
Why It Matters
If you're running text-heavy workloads on a budget, Mercury 2.5's fast speed and low cost might be a game-changer. However, for teams needing high intelligence in finance or legal tasks, sticking with higher-ranking models like Anthropic's Claude might be more prudent.
Comments
Be the first to comment
Enjoyed this article?
Get it daily. 7am. Free. Reads in 5 minutes.
Join 3,567 builders reading daily.