Unverified50% confidenceFactExact time
Luna作为轻量模型，其参数规模和推理计算量显著低于Sol，单次调用成本可能仅为Sol的几分之一
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/8/2026
First Seen
Valid until: 11/6/2026
Sources
Related Claims
VerifiedSol is the flagship model, Terra is the mid-tier balancing performance and cost, and Luna is the smallest, lowest-latency version.70% similarUnverifiedSol's output token count and completion time are both less than half of competitors', with an estimated cost approximately one-third lower.67% similarUnverified在编程智能体测试中，Sol 输出的 Token 少了一半以上，成本约为 Faber-5 的 1/366% similarUnverified轻量版模型单次推理的算力消耗往往只有旗舰版本的十分之一甚至更低61% similarUnverifiedOn Exploit Bench, Sol can compete with Anthropic models while using approximately one-third of the output tokens.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/705747API
curl https://kongchang.com/api/v1/knowledge/claims/705747MCP
get_claim(id=705747)