순위

가용성 순위 최근 24시간

최근 24시간 동안의 요청 성공률이며, 가짜 200 응답, 스트림 끊김, 형식 오류는 모두 실패로 집계합니다. 범위는 샘플 수를 바탕으로 계산한 95% Wilson 신뢰 구간이므로 샘플이 적은 경로일수록 구간이 넓습니다.

데이터경로 145개 · 중계 서비스 13곳2026. 11. 13. 5시 9분 34초 동기화, 15분마다 업데이트

모델: DeepSeek V4 ProGrok 4.6Claude Opus 5GPT 5.6 TerraGLM 5.3DeepSeek V4.1 FlashClaude Fable 5GPT 6.1 SolGemini 3.6 FlashGrok 4.7Claude Haiku 5.5Claude Fable 5.1Kimi K3Gemini 3.8 FlashGPT 6 AstraGPT 5.5Gemini 3.7 FlashQwen 3.8 MaxClaude Sonnet 5Grok 4.5GPT 5.6 SolGemini 3.1 Pro PreviewGPT 6 SolGemini 3.5 FlashClaude Opus 5.5DeepSeek V4 FlashGLM 5.3 Flash

순위	범위	모델 @ 중계 서비스	가용성	샘플	첫 토큰	세션 비용	데이터 보안
1	1–59	DeepSeek V4 Pro@ Modelflare	99.0%^±1.7	199	1.35 s	¥11.4	4
2	1–59	Grok 4.6@ Modelflare	99.0%^±1.7	197	1.16 s	¥0.28	4
3	1–59	Claude Opus 5@ Modelflare	99.0%^±1.7	195	1.21 s	¥18.3	4
4	1–59	GPT 5.6 Terra@ Modelflare	99.0%^±1.7	192	1.03 s	¥0.89	4
5	1–59	Grok 4.6@ Modelflare	98.5%^±1.9	197	6.17 s	¥3.10	4
6	1–59	GLM 5.3@ Modelflare	98.5%^±1.9	197	0.96 s	¥8.56	4
7	1–59	DeepSeek V4.1 Flash@ Modelflare	98.2%^±2.0	217	1.31 s	¥2.59	4
8	1–59	Claude Fable 5@ Modelflare	98.7%^±2.1	153	0.93 s	¥36.5	4
9	1–59	GPT 6.1 Sol@ Modelflare	98.7%^±2.2	148	0.87 s	¥0.44	4
10	1–59	Claude Fable 5@ Modelflare	98.6%^±2.2	146	1.77 s	¥46.0	4
11	1–59	Gemini 3.6 Flash@ Modelflare	98.6%^±2.2	146	1.79 s	¥3.29	4
12	1–59	Grok 4.7@ Modelflare	98.3%^±2.2	176	5.11 s	¥3.10	4
13	1–59	Claude Haiku 5.5@ Modelflare	98.0%^±2.1	200	0.94 s	—	4
14	1–59	Claude Fable 5.1@ Modelflare	98.2%^±2.2	170	1.34 s	¥4.38	4
15	1–59	Kimi K3@ Modelflare	98.0%^±2.1	198	1.16 s	¥28.6	4
16	1–59	Gemini 3.8 Flash@ Modelflare	98.6%^±2.4	139	1.40 s	¥3.29	4
17	1–59	GPT 5.6 Terra@ Modelflare	98.5%^±2.4	138	1.56 s	¥2.38	4
18	1–61	GPT 6 Astra@ Modelflare	98.5%^±2.4	135	1.33 s	¥2.19	4
19	1–61	GPT 5.5@ Modelflare	98.2%^±2.3	164	1.82 s	¥2.23	4
20	1–61	GPT 5.5@ Modelflare	98.1%^±2.3	161	1.01 s	¥1.12	4
21	1–61	Gemini 3.7 Flash@ Modelflare	98.5%^±2.5	130	1.32 s	¥3.29	4
22	1–61	GPT 5.5@ Modelflare	97.8%^±2.3	184	1.01 s	¥5.95	4
23	1–61	Qwen 3.8 Max@ Modelflare	98.0%^±2.6	147	1.06 s	¥20.4	4
24	1–61	Claude Sonnet 5@ Modelflare	97.7%^±2.5	172	1.58 s	¥9.20	4
25	1–61	Grok 4.5@ Modelflare	97.2%^±2.3	215	5.20 s	¥3.10	4
26	1–61	Claude Opus 5@ Modelflare	97.6%^±2.5	166	1.44 s	¥23.0	4
27	1–61	GPT 5.6 Terra@ Modelflare	97.9%^±2.7	141	1.04 s	¥0.45	4
28	1–61	GPT 5.6 Sol@ Modelflare	97.8%^±2.7	138	1.34 s	¥37.2	4
29	1–61	GPT 5.6 Sol@ Modelflare	97.8%^±2.8	134	1.84 s	¥5.95	4
30	1–61	Gemini 3.1 Pro Preview@ Modelflare	98.2%^±2.9	110	0.91 s	¥4.46	4
31	1–61	Claude Opus 5@ Modelflare	97.2%^±2.6	177	0.99 s	¥2.19	4
32	1–61	GPT 6 Sol@ Modelflare	97.7%^±2.8	132	1.21 s	¥0.44	4
33	1–61	Claude Sonnet 5@ Modelflare	97.7%^±2.9	130	1.77 s	¥7.30	4
34	1–61	GPT 5.6 Sol@ Modelflare	98.1%^±3.1	105	1.24 s	¥2.23	4
35	1–61	Claude Fable 5@ Modelflare	98.1%^±3.1	104	1.32 s	¥4.38	4
36	1–61	Grok 4.5@ Modelflare	97.6%^±3.0	126	1.84 s	¥0.28	4
37	1–61	Gemini 3.5 Flash@ Modelflare	98.1%^±3.1	103	1.78 s	¥3.35	4
38	1–61	Claude Opus 5.5@ Modelflare	97.3%^±2.9	146	1.57 s	¥1.75	4
39	1–61	DeepSeek V4 Flash@ Modelflare	97.2%^±2.9	145	1.72 s	¥2.59	4
40	1–61	Claude Sonnet 5@ Modelflare	97.2%^±2.9	143	1.29 s	¥0.88	4
41	1–61	GPT 5.5@ Modelflare	97.5%^±3.2	118	0.93 s	¥37.2	4
42	1–61	GPT 5.6 Terra@ Modelflare	97.3%^±3.4	111	0.85 s	¥14.9	4
43	1–61	GPT 5.6 Sol@ Modelflare	97.1%^±3.6	102	1.31 s	¥1.12	4
44	1–63	GPT 6 Astra@ 智流 FluxLane	100.0%^±5.5	31	9.76 s	—	4
45	1–63	GPT 6.1 Sol@ 智流 FluxLane	100.0%^±5.5	31	8.85 s	—	4
46	1–63	DeepSeek V4 Pro@ 智流 FluxLane	100.0%^±5.5	31	2.27 s	—	4
47	1–63	Claude Sonnet 5@ 智流 FluxLane	100.0%^±5.5	31	23.19 s	—	4
48	1–63	Claude Sonnet 5@ 智流 FluxLane	100.0%^±5.5	31	23.19 s	—	4
49	1–63	Claude Sonnet 5@ 智流 FluxLane	100.0%^±5.5	31	23.19 s	—	4
50	1–63	GPT 5.6 Sol@ 智流 FluxLane	100.0%^±5.5	31	7.46 s	—	4
51	1–64	Claude Fable 5.1@ 智流 FluxLane	96.7%^±8.0	30	7.00 s	—	4
52	1–64	GPT 5.6 Terra@ 智流 FluxLane	96.7%^±8.0	30	12.41 s	—	4
53	1–64	Claude Opus 5.5@ 智流 FluxLane	93.3%^±9.7	30	9.05 s	—	4
54	1–64	Claude Fable 5@ 智流 FluxLane	93.1%^±10.0	29	6.68 s	—	4
55	1–64	Claude Fable 5@ 智流 FluxLane	93.1%^±10.0	29	6.68 s	—	4
56	1–64	Claude Fable 5@ 智流 FluxLane	93.1%^±10.0	29	6.68 s	—	4
57	1–64	Claude Opus 5@ 智流 FluxLane	90.0%^±11.1	30	6.57 s	—	4
58	1–64	Claude Opus 5@ 智流 FluxLane	90.0%^±11.1	30	6.57 s	—	4
59	1–64	Claude Opus 5@ 智流 FluxLane	90.0%^±11.1	30	6.57 s	—	4
60	18–64	Grok 4.5@ 智流 FluxLane	87.1%^±11.9	31	4.79 s	—	4
61	18–64	Grok 4.5@ 智流 FluxLane	87.1%^±11.9	31	4.79 s	—	4
62	44–64	Grok 4.6@ 智流 FluxLane	80.7%^±13.5	31	2.57 s	—	4
63	44–64	Grok 4.6@ 智流 FluxLane	80.7%^±13.5	31	2.57 s	—	4
64	51–64	DeepSeek V4.1 Flash@ 智流 FluxLane	73.3%^±15.1	30	1.83 s	—	4
65	65–66	Claude Fable 5@ CheapAI	100.0%^±5.5	31	5.01 s	—	3
66	65–70	GPT 5.6 Sol@ CheapAI	80.7%^±13.5	31	18.77 s	¥0.67	3
67	66–70	Claude Opus 5.5@ CheapAI	77.4%^±14.2	31	19.67 s	—	3
68	66–70	Kimi K3@ CheapAI	77.4%^±14.2	31	3.02 s	¥4.40	3
69	66–70	Claude Opus 5@ CheapAI	74.2%^±14.8	31	16.57 s	—	3
70	66–70	GPT 5.6 Terra@ CheapAI	64.5%^±16.0	31	12.36 s	¥0.27	3
71	71–133	DeepSeek V4.1 Flash@ MuskAI	100.0%^±5.5	31	3.63 s	—	0
72	71–133	Grok 4.7@ MuskAI	100.0%^±5.5	31	12.91 s	—	0
73	71–133	Claude Sonnet 5@ MuskAI	100.0%^±5.5	31	3.52 s	—	0
74	71–133	DeepSeek V4.1 Flash@ ZNB.ai	100.0%^±5.7	30	2.51 s	—	0
75	71–133	GPT 6 Astra@ MuskAI	100.0%^±5.7	30	6.04 s	—	0
76	71–133	GPT 6 Astra@ NexoToken	100.0%^±5.7	30	8.02 s	—	0
77	71–133	GPT 6.1 Sol@ Zashmodel	100.0%^±5.7	30	4.62 s	¥2.38	0
78	71–133	GPT 5.6 Sol@ NexoToken	100.0%^±5.7	30	9.64 s	—	0
79	71–133	GPT 5.6 Terra@ NexoToken	100.0%^±5.7	30	12.37 s	—	0
80	71–133	GPT 5.6 Terra@ NexoToken	100.0%^±5.7	30	12.90 s	—	0
81	71–133	Claude Opus 5@ NexoToken	100.0%^±5.7	30	6.45 s	—	0
82	71–133	GPT 6 Sol@ NexoToken	100.0%^±5.8	29	10.13 s	—	0
83	71–133	GPT 5.6 Sol@ MuskAI	96.8%^±7.8	31	5.11 s	—	0
84	71–133	GLM 5.3 Flash@ MuskAI	96.8%^±7.8	31	11.76 s	—	0
85	71–133	GPT 6 Astra@ ZNB.ai	96.7%^±8.0	30	5.20 s	—	0
86	71–133	Grok 4.7@ NexoToken	96.7%^±8.0	30	10.99 s	—	0
87	71–133	Kimi K3@ NexoToken	96.7%^±8.0	30	4.34 s	—	0
88	71–133	DeepSeek V4 Pro@ NexoToken	96.7%^±8.0	30	9.57 s	—	0
89	71–133	Claude Sonnet 5@ NexoToken	96.7%^±8.0	30	6.55 s	—	0
90	71–133	Grok 4.5@ NexoToken	96.7%^±8.0	30	18.80 s	—	0
91	71–133	Claude Fable 5.1@ Apiporter	100.0%^±8.8	18	2.17 s	—	0
92	71–133	DeepSeek V4.1 Flash@ IKunCode	100.0%^±8.8	18	3.95 s	—	0
93	71–133	DeepSeek V4.1 Flash@ CheapRouter	100.0%^±8.8	18	4.17 s	¥1.30	0
94	71–133	GPT 6 Astra@ Apiporter	100.0%^±8.8	18	2.88 s	—	0
95	71–133	GPT 6 Astra@ IKunCode	100.0%^±8.8	18	3.49 s	¥15.4	0
96	71–133	GPT 6 Astra@ CheapRouter	100.0%^±8.8	18	4.04 s	¥3.30	0
97	71–133	GPT 6 Astra@ A6api	100.0%^±8.8	18	9.99 s	¥2.04	0
98	71–133	Claude Opus 5.5@ Apiporter	100.0%^±8.8	18	2.03 s	—	0
99	71–133	Claude Opus 5.5@ CheapRouter	100.0%^±8.8	18	3.97 s	¥4.40	0
100	71–133	GPT 6.1 Sol@ CheapRouter	100.0%^±8.8	18	7.18 s	¥0.66	0
101	71–133	GPT 6.1 Sol@ Apiporter	100.0%^±8.8	18	2.45 s	—	0
102	71–133	Grok 4.7@ IKunCode	100.0%^±8.8	18	4.88 s	¥1.27	0
103	71–133	Grok 4.7@ Cloudbridge	100.0%^±8.8	18	3.31 s	—	0
104	71–133	Claude Sonnet 5@ IKunCode	100.0%^±8.8	18	2.88 s	¥4.40	0
105	71–133	GPT 5.6 Sol@ 境Ai	100.0%^±8.8	18	3.32 s	¥4.48	0
106	71–133	Grok 4.6@ IKunCode	100.0%^±8.8	18	4.13 s	¥1.27	0
107	71–133	Grok 4.5@ IKunCode	100.0%^±8.8	18	4.78 s	¥1.27	0
108	71–134	GPT 6 Astra@ Zashmodel	93.3%^±9.7	30	8.16 s	¥11.9	0
109	71–134	Grok 4.6@ NexoToken	93.3%^±9.7	30	12.86 s	—	0
110	71–135	GPT 6.1 Sol@ NexoToken	90.0%^±11.1	30	6.70 s	—	0
111	71–135	GPT 5.6 Sol@ NexoToken	90.0%^±11.1	30	6.39 s	—	0
112	71–135	GPT 5.6 Sol@ NexoToken	90.0%^±11.1	30	4.74 s	—	0
113	71–135	DeepSeek V4 Flash@ NexoToken	90.0%^±11.1	30	13.78 s	—	0
114	71–135	GLM 5.3@ NexoToken	90.0%^±11.1	30	7.27 s	—	0
115	71–135	DeepSeek V4.1 Flash@ Cloudbridge	94.4%^±12.4	18	6.48 s	—	0
116	71–135	Claude Opus 5@ IKunCode	94.4%^±12.4	18	6.04 s	¥11.0	0
117	71–135	DeepSeek V4 Flash@ A6api	94.4%^±12.4	18	6.41 s	¥2.04	0
118	71–135	DeepSeek V4.1 Flash@ NexoToken	89.7%^±11.4	29	6.03 s	—	0
119	71–135	GPT 5.6 Sol@ NexoToken	89.7%^±11.4	29	11.22 s	—	0
120	71–135	GPT 6.1 Sol@ ZNB.ai	86.7%^±12.2	30	4.67 s	—	0
121	71–135	GPT 6 Sol@ A6api	88.9%^±14.8	18	8.05 s	¥2.04	0
122	71–135	Claude Fable 5.1@ A6api	88.9%^±14.8	18	24.58 s	¥2.04	0
123	71–135	Claude Opus 5.5@ A6api	88.9%^±14.8	18	15.94 s	¥2.04	0
124	71–135	Kimi K3@ A6api	88.9%^±14.8	18	5.53 s	¥2.04	0
125	71–135	DeepSeek V4 Pro@ A6api	88.9%^±14.8	18	6.48 s	¥2.04	0
126	71–135	Claude Fable 5@ NexoToken	80.0%^±13.9	30	7.25 s	—	0
127	71–135	GLM 5.3 Flash@ NexoToken	80.0%^±13.9	30	9.14 s	—	0
128	71–135	Claude Opus 5.5@ ZNB.ai	79.3%^±14.3	29	21.85 s	—	0
129	71–135	Claude Opus 5.5@ IKunCode	83.3%^±16.7	18	12.04 s	¥8.80	0
130	71–135	Claude Fable 5@ A6api	83.3%^±16.7	18	19.34 s	¥2.04	0
131	71–135	Claude Opus 5@ A6api	83.3%^±16.7	18	11.82 s	¥2.04	0
132	71–135	Claude Fable 5.1@ Cloudbridge	77.8%^±18.1	18	14.16 s	—	0
133	71–135	Claude Sonnet 5@ A6api	77.8%^±18.1	18	9.30 s	¥2.04	0
134	110–135	Claude Opus 5.5@ Zashmodel	60.0%^±16.5	30	2.38 s	¥4.77	0
135	108–135	Claude Opus 5.5@ Cloudbridge	61.1%^±20.5	18	12.08 s	—	0
136	136–145	Claude Fable 5@ 境Ai	5.6%^±12.4	18	1.95 s	—	0
137	136–145	Gemini 3.8 Flash@ MuskAI	0.0%^±5.5	31	—	—	0
138	136–145	GPT 6 Sol@ ZNB.ai	0.0%^±5.7	30	—	—	0
139	136–145	GPT 5.6 Sol@ NexoToken	0.0%^±5.7	30	—	—	0
140	136–145	GPT 5.6 Sol@ ZNB.ai	0.0%^±5.7	30	—	—	0
141	136–145	GPT 5.6 Sol@ ZNB.ai	0.0%^±5.7	30	—	—	0
142	136–145	Grok 4.7@ 境Ai	0.0%^±8.8	18	—	—	0
143	136–145	Grok 4.6@ 境Ai	0.0%^±8.8	18	—	—	0
144	136–145	GPT 5.5@ 境Ai	0.0%^±8.8	18	—	¥4.48	0
145	136–145	Grok 4.5@ 境Ai	0.0%^±8.8	18	—	—	0

자주 묻는 질문

가용성 순위는 어떻게 계산합니까?

“범위”는 무엇을 의미합니까?

각 경로의 샘플 수로 95% 신뢰 구간을 계산한 뒤, 어떤 경로가 뚜렷하게 더 좋거나 나쁜지를 따져 가능한 순위 범위를 구합니다. 가용성이 100%라도 요청이 몇 번뿐인 경로는 범위가 매우 넓습니다.

같은 중계 서비스가 왜 여러 번 나옵니까?

각 행은 하나의 “모델 @ 중계 서비스” 경로입니다. 같은 중계 서비스라도 모델과 그룹에 따라 성능이 크게 다를 수 있으므로 따로 순위를 매깁니다.