ClickHouse @ClickHouseDB · 01 Sep 2026
We benchmarked 28 LLMs on agentic analytics: 201 real questions from our internal analytics agent, replayed against a full synthetic reconstruction of our data warehouse. @claudeai Fable 5.1 holds the top spot at 76.6% correctness. Thanks to @AnthropicAI for letting us https://t.co/srPkGy3qDt
2 965Views
36Likes
3Reposts
3Replies
0Quotes
15Bookmarks
Is that a lot?
1.25×vs this author's median2 370 views is typical
60Percentile for this authorof 5 recent posts
1.02×vs 10K–100K median2 896 views is typical
15.02%Reachviews ÷ followers
1.42%Engagement rateof viewers reacted