Unverified50% confidenceFactExact time
Lightweight models such as GPT-4o mini and Claude Haiku remain low-priced or covered by subscriptions, while flagship models are gradually migrating toward pay-as-you-go or premium subscription tiers.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/24/2026
First Seen
Valid until: 10/22/2026
Sources
Anthropic放弃订阅制转向按量计费，是明智还是自断后路？
redditr/ClaudeAI7/12/2026
Related Claims
UnverifiedClaude Haiku 4.5 still holds a clear pricing advantage over peer-level models such as GPT-4o mini.83% similarUnverifiedFlagship models (such as GPT-4o and Claude 3.5 Sonnet) and lightweight models (such as GPT-4o-mini and Claude Haiku) can differ in price by 10–20x, yet the performance gap on a large number of routine tasks is not significant.81% similarUnverifiedGPT-4o mini、Claude Haiku等轻量模型采用蒸馏技术，在参数量缩减80%以上情况下仍能保留旗舰模型60-70%的综合能力80% similarUnverifiedThe API pricing gap between GPT-4o mini and GPT-4o is approximately 10 to 20 times, and the price difference between Claude Haiku and Opus is on the same order of magnitude.79% similarVerifiedClaude Haiku and GPT-4o mini are classified as lightweight models with fast inference speeds and low API call costs, suitable for tasks such as format conversion and simple refactoring.77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/607282API
curl https://kongchang.com/api/v1/knowledge/claims/607282MCP
get_claim(id=607282)