Public platform API
Query models, prices, service quality, and platform status without authentication.
These endpoints are free and require no authentication. Use them for model and price comparisons, monitoring, or display integrations. Base URL: https://goodputai.cn.
Model catalog and pricing
curl https://goodputai.cn/api/v1/public/modelsReturns all models on sale and their current prices:
{
"models": [{
"public_model": "deepseek/deepseek-v4-flash",
"display_name": "DeepSeek: DeepSeek V4 Flash",
"channel_type": "partner",
"context_window": 1000000,
"max_output_tokens": 384000,
"modalities": ["text"],
"capabilities": ["tools", "reasoning", "caching", "structured"],
"price": {
"input_per_1m": "1.000000",
"output_per_1m": "2.000000",
"cache_read_per_1m": "0.020000",
"windows": [{
"name": "peak-am", "tz": "Asia/Shanghai",
"weekday_mask": 127, "start_minute": 540, "end_minute": 720,
"input_per_1m": "2.000000", "output_per_1m": "4.000000"
}],
"currency": "CNY", "unit": "1M_tokens"
}
}, {
"public_model": "wan/wan2.6-t2v",
"display_name": "Wan: Wan2.6 T2V",
"channel_type": "self",
"modalities": ["video"],
"price": {
"mode": "meter",
"per_video_second_tiers": {"720p": "0.600000", "1080p": "1.000000"},
"currency": "CNY"
}
}],
"generated_at": "..."
}Fields:
| Field | Description |
|---|---|
public_model | Model ID used in requests |
channel_type | self: platform-operated infrastructure; partner: official access |
capabilities | tools / reasoning / caching / structured capability labels |
price.windows[] | Time-based prices. weekday_mask is a Monday-to-Sunday bitmask (127 = every day). start_minute / end_minute are minutes from midnight in the specified local time zone. Outside these windows, top-level prices apply |
price.tiers[] | Long-context tiers for the standard time window. When the request’s full input, including cache reads and writes, reaches from_input_tokens, all four dimensions of the whole request use that tier’s prices |
price.windows[].tiers[] | Tiers within each time window; the final price matrix when time-based and context-based pricing combine |
price.*_per_1m | ¥/million tokens; null means the dimension has no separate price |
price.per_image | Image-generation price, ¥/image. For mode: meter, the applicable metered unit has a value while token prices are all null |
price.per_1k_characters | Speech-synthesis price, ¥/thousand characters |
price.per_video_second | Video price, ¥/second, for models without resolution tiers |
price.per_video_second_tiers | Resolution-to-price map (tier → ¥/second), such as {"720p": "0.600000", "1080p": "1.000000"}. When present, per_video_second is null |
Service quality for one model (SLO)
curl "https://goodputai.cn/api/v1/public/models/slo?model=z-ai/glm-5.2"Returns overall availability, P95 latency, and an hourly availability timeline for the last 24 hours, together with each supply channel’s availability, latency, and output throughput. Channel identities are anonymized as “Channel A/B”.
Live platform status
curl https://goodputai.cn/api/v1/public/platform/sloReturns total requests, availability, mean time to first token (ttft_mean_ms), and output throughput (output_tps) for the last 24 hours.
Platform usage statistics
curl https://goodputai.cn/api/v1/public/platform/statsReturns the platform’s month-to-date token total (month_tokens).
These responses use short caches of approximately 1–5 minutes. They are suitable for polling; do not use them for high-frequency load tests.