Unverified50% confidenceBenchmarkExact time
根据Artificial Analysis和Cursor的benchmark，GPT-5.6 Sol完成任务约需17k-28k token，而Anthropic的Fable需要36k至103k token
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/15/2026
First Seen
Valid until: 12/14/2026
Sources
Related Entities
Related Claims
Unverified根据Artificial Analysis数据，Fable完成单个任务约用33k token，Opus 5约用37k token77% similarUnverified在Gemma 4 12B模型的Token生成测试中，Spark达26.4 tokens/s、Halo 24.6 tokens/s、M4 Pro最快33.8 tokens/s66% similarVerifiedClaude 3.5 Sonnet支持200K token上下文，GPT-4 Turbo支持100K token上下文66% similarUnverifiedOpus 4.7每次解题的中位输出token数为60,000，而GPT 5.5仅需16,00066% similarUnverified在Artificial Analysis的Omniscience幻觉指标上，GPT-5.6 Sol Max高达89%，高于Claude Fable的55%和GLM 5.2的28%64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/923131API
curl https://kongchang.com/api/v1/knowledge/claims/923131MCP
get_claim(id=923131)