[KongchangAI]
Unverified50% confidenceFactExact time

The per-token price differs significantly across Claude models (such as Claude 3 Haiku, Sonnet and Opus), and output tokens are usually more expensive than input tokens.

1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/30/2026
First Seen
Valid until: 1/28/2027

Sources

Claude Usage and Cost Analytics API: Attribute Spend by API Key and User

Related Entities

Claude 3 HaikumodelClaude 3 SonnetmodelClaude 3 Opusmodel

Related Claims

UnverifiedThe X-high thinking tier of Qwen3.8 27B consumes an extremely large number of tokens (e.g., 11,000 for drawing a pelican and 35,000 for drawing a dragon), while the medium tier is the best cost-performance sweet spot.56% similar

Cite This Claim

Stable URI
https://kongchang.com/claim/1113595
API
curl https://kongchang.com/api/v1/knowledge/claims/1113595
MCP
get_claim(id=1113595)