/

Data 2026-10-09 04:41 · USD/CNY 6.74

← Sinan Lab › Monthly report › 2026-10

Sinan Lab · Monthly · 2026-10-01 to 2026-10-09 (rolling update)

China Model API Relay Market Monthly · 2026 年 10 月

One issue per month. Figures come from daily automated measurement of 1966 confirmed relay sites; judgments and analysis are written by Sinan Lab, based only on these measurements, and are not a recommendation of any channel.

Not a cent from those we measureNo paid listing, paid verification or paid fast-trackRankings are not for saleEvery board is sorted by the measured value; no commercial variableNo referral parametersOutbound links only count clicks; no commissions

confirmed relay sites

1966

Subscription (Sub2API family) 797 · Pay-as-you-go (one-api family) 964

Foreign closed models · relay median ratio

22%

9 models; Chinese-model median 85% (14 models)

Index level

87.3 → 86.4

Discount ratio snapshot 58% → 59% (affected by sample composition)

Mainstream model price changes

116 changes

Up 65 · Down 51; domestic down 15 up 29

Our reading of the month

本期分析撰写中。

一、规模与结构

Panel family	Sites	Open reg.	Close	Unknown	With public quotes	Median ratio
one-api	871	681	97	93	367	28%
sub2api	797	648	86	63	0	—
multimodal-aggregator	152	45	13	94	0	—
one-api/new-api	90	60	14	16	35	70%
openai-compatible	42	22	2	18	0	—
multi-vendor	6	3	0	3	1	3%

Panel family is determined by public API fingerprint; Sub2API-family plan prices sit behind login, so there is no public quote this issue.

Login method	one-api family	Sub2API family
Email	618	787
GitHub	230	118
LINUX DO	129	44
Google	0	90
OIDC	38	0
Discord	25	0
Telegram	16	0
WeChat	10	0

TLD	Sites
.com	591
.ai	192
.top	177
.cc	147
.xyz	105
.cn	94
.net	65
.vip	38

Source	Sites this period
content_fp	48
adjacency	10
github	6
ct_log+mass_probe	6
feeds	3
Directory: batch	2

二、价格

Whole marketFlagshipMidFast

Model	Vendor	Tier	Official $/M	Relay median $/M	Ratio	Sites
GPT-6.1 Sol	openai	Mid	10.00	1.48	15%	134
GPT-6 Astra	openai	Flagship	50.00	7.42	15%	209
GPT-6 Sol	openai	Mid	10.00	1.48	15%	155
Grok 4.7	xai	Mid	6.00	0.891	15%	57
Claude Opus 5.5	anthropic	Flagship	20.00	4.36	22%	98
Claude Fable 5.1	anthropic	Flagship	50.00	11.87	24%	87
Gemini 3.8 Flash	google	Mid	3.75	1.11	30%	70
Claude Sonnet 5.5	anthropic	Mid	10.00	3.34	33%	68
GPT-6 Luna	openai	Fast	0.500	0.178	36%	89
GLM-5.2	zhipu	Mid	7.20	3.33	46%	120
DeepSeek V4.1 Flash	deepseek	Fast	1.20	0.594	49%	64
GLM-5.3	zhipu	Mid	7.00	4.16	59%	87
GLM-5.3 Flashx	zhipu	Fast	1.25	0.753	60%	20
GLM-5.3 Flash	zhipu	Fast	0.500	0.327	65%	74
Kimi K3	moonshot	Mid	15.00	12.57	84%	106
Qwen 3.8 Max	alibaba	Mid	6.00	5.08	85%	92
Qwen 3.8 Flash	alibaba	Fast	0.470	0.401	85%	36
MiniMax M3	minimax	Fast	1.20	1.02	85%	87
Qwen 3.7 Plus	alibaba	Fast	1.28	1.19	93%	72
DeepSeek V4 Flash	deepseek	Fast	0.668	0.668	100%	141
Qwen 3.7 Max	alibaba	Mid	4.42	5.08	115%	85
Qwen 3.7 Flash	alibaba	Fast	0.130	0.658	506%	18
DeepSeek V4 Pro	deepseek	Fast	0.418	2.61	623%	139

Discount ratio = relay market median effective price ÷ official reference price; domestic / overseas by the vendor's country of registration.

Tier	Quotes	Below cost floor	Below bulk discount	Within range	Near public	Above public	Far above
Flagship	435	51%	17%	6%	20%	2%	4%
Mid	1087	31%	13%	11%	30%	7%	8%
Fast	836	11%	10%	16%	23%	14%	27%

Band definitions are in section 3 of the definitions page: <15% below cost floor · 15–40% below common bulk discount · 40–80% within range · 80–125% near public price · 125–200% above · >200% well above.

Top-up rate (CNY per $1 nominal)	Sites	Share
1	201	51%
7.3	74	19%
7	18	5%
6.8	11	3%
0.8	8	2%
0.1	7	2%
0.2	7	2%
0.5	5	1%

Each site's most common top-up rate; 393 sites with public quotes. 1.0 = ¥1 buys $1 of quota (effective price about 15% of official); 6.8–7.3 ≈ real exchange rate 1:1.

Largest moves in market median this period

Model	Start → end $/M	Change	Sites · % of official
Qwen 3.7 Flash Fast	$1.19 → $0.658	-45%	18 sites · 506%
Claude Fable 5.1 Flagship	$13.12 → $11.87	-10%	87 sites · 24%
Claude Opus 5.5 Flagship	$4.54 → $4.36	-4%	98 sites · 22%
GPT-6.1 Sol Mid	$1.48 → $1.48	+0%	134 sites · 15%
GPT-6 Astra Flagship	$7.42 → $7.42	+0%	209 sites · 15%
Qwen 3.7 Max Mid	$4.28 → $5.08	+19%	85 sites · 115%
Kimi K3 Mid	$11.58 → $12.57	+9%	106 sites · 84%
Claude Sonnet 5.5 Mid	$3.12 → $3.34	+7%	68 sites · 33%
DeepSeek V4 Pro Fast	$2.47 → $2.61	+6%	139 sites · 623%
GLM-5.3 Flashx Fast	$0.727 → $0.753	+4%	20 sites · 60%

3. Reachability and probes

Eligible sites	7-day 100%	99%–99.9%	Below 99%	Our outage rounds excluded
282	76	73	133	—

Lowest reachability	7-day reach	Probes	Handshake p50
api520.pro	57.1%	175	4949ms
opnai.top	64.6%	175	3168ms
suanlitaikong.com	65.7%	175	4798ms
gvmz.systems	67.1%	173	2931ms
fishxcode.com	90.2%	174	3685ms

Probe	Site×model pairs	Consistent	Differs	Capability pairs	Below median
This period	39	0	0	9	0

Audit · hold reason	Count
Leader gap	4
Lone outliers	8
Unit hints	31
unit_suspect	209
Released this period	0

4. Image and video

Video family	Sites	Comparable quotes	Official ref $/s	Family median $/s	Median ÷ ref	Below discount / above public
Wan (Alibaba)	48	192	$0.045	$0.080	179%	14 / 79
Veo（Google）	33	142	$0.100	$0.065	65%	97 / 2
Kling (Kuaishou)	54	103	$0.084	$0.059	71%	28 / 21
Seedance (ByteDance)	29	90	$0.099	$0.083	84%	29 / 26
Hailuo / H3（MiniMax）	57	70	$0.080	$0.049	62%	14 / 17
Grok Imagine Video (xAI)	46	52	$0.050	$0.020	40%	31 / 5
Vidu (Shengshu)	40	46	$0.042	$0.080	191%	2 / 37

Image family	Sites	Comparable quotes	Official ref $/image	Family median $/image	Median ÷ ref	Below discount / above public
Nano Banana / Gemini Image（Google）	84	218	$0.034	$0.029	86%	86 / 35
Seedream (ByteDance)	62	195	$0.033	$0.030	91%	27 / 31
Qwen-Image (Alibaba)	58	138	$0.027	$0.039	144%	2 / 82
Grok Imagine Image (xAI)	73	100	$0.020	$0.021	106%	32 / 31
Wan Image (Alibaba)	43	36	$0.074	$0.082	110%	0 / 18
Kling Image (Kuaishou)	6	23	$0.028	$0.006	21%	14 / 5
FLUX（BFL）	18	4	$0.100	$0.030	30%	0 / 0

The reference price uses the lowest official tier (resolution / duration); relay quotes are converted to the same tier; families without a public official price get no ratio.

5. Compute cost layer

GPU	Vast.ai on-demand median $/h	RunPod secure $/h
RTX 4090	0.46	0.89
RTX 4090 D	—	—
RTX 5090	0.61	1.19
RTX 5090 D	—	—
RTX A6000	0.42	0.59
RTX PRO 6000	—	—
L40S	0.74	1.09
V100 32GB	—	—

Open model	Official $/M	Relay median $/M	Ratio	Whether the official price is below the H100 single-card cost line (about $0.185/M)
Qwen 3.7 Flash	0.130	0.658	506%	Yes
DeepSeek V4 Pro	0.418	2.61	623%	No
Qwen 3.8 Flash	0.470	0.401	85%	No
GLM-5.3 Flash	0.500	0.327	65%	No
DeepSeek V4 Flash	0.668	0.668	100%	No
DeepSeek V4.1 Flash	1.20	0.594	49%	No
MiniMax M3	1.20	1.02	85%	No
GLM-5.3 Flashx	1.25	0.753	60%	No
Qwen 3.7 Plus	1.28	1.19	93%	No
Qwen 3.7 Max	4.42	5.08	115%	No
Qwen 3.8 Max	6.00	5.08	85%	No
GLM-5.3	7.00	4.16	59%	No
GLM-5.2	7.20	3.33	46%	No
Kimi K3	15.00	12.57	84%	No

The cost line is a rough estimate under the assumptions in section 13 of the methodology doc (H100 $2.00/h, 3000 tokens/s), for order of magnitude only.

Definitions and corrections

All metric definitions are in Definitions; raw data:2026-10.json、price-index.json、data_v2.json. Figures are recalculated daily during the month, finalized at month end and marked "final"; the analysis text is updated together with the final version.

Cite this page

Data may be freely cited and republished with attribution and a link. We take no money from measured channels, sell no rankings and add no referral parameters, so you can cite us as an independent third-party measurement.

司南实验室，《中国模型 API 中转市场月报 · 2026 年 10 月》，2026-10-09，https://compute.sinanlab.com/report/2026-10（访问日期 ____）

Sinan Lab, "中国模型 API 中转市场月报 · 2026 年 10 月", 2026-10-09, https://compute.sinanlab.com/report/2026-10 (accessed ____)

Media, research institutions and financial institutions needing raw data or custom definitions: write to [email protected], see Press & researchers。

Clear direction, grounded judgement.