- alazyz2026-09-2439789
token 消耗量咋样,我看 Claude code 都把默认模型改成 Opus5.5 了
主楼 - 大鲨鱼2026-09-246550
更耐用了,因为成本低了
- alazyz2026-09-2439789
- Boots2026-09-2416245
高强度用过FABLE 5.1 ,没啥感觉
- alazyz2026-09-2439789
- 昆吾断鸿2026-09-2453518
Total cost: $694.26
Total duration (API): 10h 58m 3s
Total duration (wall): 1d 1h 20m
Total code changes: 34443 lines added, 2477 lines removed
Usage by model:
claude-opus-5: 65.1k input, 947.2k output, 657.8m cache read, 2.1m cache write ($371.93)
claude-haiku-4-5: 163.1k input, 4.8k output, 0 cache read, 0 cache write ($0.1870)
claude-fable-5-1: 10.8k input, 216.9k output, 127.1m cache read, 2.3m cache write ($89.50)
claude-opus-5-5: 1.0m input, 1.9m output, 748.3m cache read, 8.1m cache write ($232.64)
Prompt cache (main): 9 requests · 89% of input tokens from cache · 1 miss (last 41m 14s ago — likely cause: idle past the 1h TTL, 231.0k tokens re-cached) · warm (1h TTL, last activity 31m 57s ago)
- neutronc2026-09-244491
- 昆吾断鸿2026-09-2453518
差不多是7d的50%的份额 我是5x
- alazyz2026-09-2439789
- neutronc2026-09-244491
- alazyz2026-09-2439789
这倒是赞同
- wyk3358585752026-09-2413521
Opus5.5 比Opus5强,计费还便宜
- Frank62026-09-2452406
真的耐用,提升很大感觉,杀死比赛了,现在反过来 gpt 不耐用了
- alazyz2026-09-2439789
我也有这种感觉,感觉是因为命中高了省下来的
- Frank62026-09-2452406
还有一点,5.5 速度快了不少,这个对比 5.1 提升也挺大的,luna 很便宜但是根本忍不了那个速度
- alazyz2026-09-2439789
看来确实不错
- kyrie-09282026-09-244893
确实耐用了
- oxy2026-09-2421177
确实耐用很多