Unverified50% confidenceBenchmarkExact time
In the intelligence range of 10 to 20 points, O1 Preview initially cost $15 per task; a few months later, Grok 3 Mini achieved a higher score at approximately one-hundredth of the price, representing a decline of nearly 99% within four months.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/12/2026
First Seen
Valid until: 1/10/2027
Sources
Related Entities
Related Claims
UnverifiedRunning the same benchmark suite, Grok 4.6 costs $1.86, Astra costs $3.26, and Grok 4.7 costs as much as $3.74 per task.64% similarUnverifiedThe total cost for all three tasks was approximately $0.59, including all retries (the video was attempted three or four times, and the briefing was re-run two or three times).63% similarUnverifiedIn the official egg-protection demonstration, using Haiku 5.5 with 10 sub-agents tried 86 designs in 58 seconds at a cost of $0.14, while Opus 5.5 running alone took 3 minutes and 37 seconds, tried 25 designs, and cost $0.47.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1004412API
curl https://kongchang.com/api/v1/knowledge/claims/1004412MCP
get_claim(id=1004412)