Hy3
tencent/hy3
Compare with Hy3 preview, Hy-MT2-1.8B, Hy-MT2-30B-A3B
CompareHy3
| State | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Healthy | 100% | 99.74% | 99.29% | 99.29% | 1.80s | 20.9s | 115 tps | 262K | $0.200 / $0.800 | |
| Healthy | 100% | 99.47% | 99.14% | 99.14% | 895ms | 3.93s | 47 tps | 262K | $0.132 / $0.528 | |
| Healthy | 100% | 95.57% | 97.43% | 97.43% | 563ms | 13.6s | 63 tps | 262K | $0.140 / $0.580 | |
| Healthy | 100% | 98.35% | 96.89% | 96.89% | 2.23s | 17.3s | 54 tps | 262K | $0.126 / $0.522 | |
Novitanovita | Healthy | 100% | 98.95% | 98.91% | 98.91% | 2.05s | 38.1s | 92 tps | 262K | $0.140 / $0.580 |
| Healthy | 100% | 99.86% | 99.66% | 99.66% | 2.54s | 14.2s | 55 tps | 262K | $0.132 / $0.528 |
- AtlasCloud100%Healthy24h 99.74%
- 7d
- 99.29%
- 30d
- 99.29%
- p50
- 1.80s
- p99
- 20.9s
- Throughput
- 115 tps
- Context
- 262K
- $/Mtok
- $0.200 / $0.800
atlas-cloud/fp8fp8 - Baidu100%Healthy24h 99.47%
- 7d
- 99.14%
- 30d
- 99.14%
- p50
- 895ms
- p99
- 3.93s
- Throughput
- 47 tps
- Context
- 262K
- $/Mtok
- $0.132 / $0.528
baidu/fp8fp8 - DeepInfra100%Healthy24h 95.57%
- 7d
- 97.43%
- 30d
- 97.43%
- p50
- 563ms
- p99
- 13.6s
- Throughput
- 63 tps
- Context
- 262K
- $/Mtok
- $0.140 / $0.580
deepinfra/fp8fp8 - GMICloud100%Healthy24h 98.35%
- 7d
- 96.89%
- 30d
- 96.89%
- p50
- 2.23s
- p99
- 17.3s
- Throughput
- 54 tps
- Context
- 262K
- $/Mtok
- $0.126 / $0.522
gmicloud/bf16bf16 - Novita100%Healthy24h 98.95%
- 7d
- 98.91%
- 30d
- 98.91%
- p50
- 2.05s
- p99
- 38.1s
- Throughput
- 92 tps
- Context
- 262K
- $/Mtok
- $0.140 / $0.580
novita - Tencent100%Healthy24h 99.86%
- 7d
- 99.66%
- 30d
- 99.66%
- p50
- 2.54s
- p99
- 14.2s
- Throughput
- 55 tps
- Context
- 262K
- $/Mtok
- $0.132 / $0.528
tencent/fp8fp8
Measured 3m ago. Latency and throughput are OpenRouter's rolling 30-minute windows.
Price changes
- Hy3Baidubaidu/fp8$0.125 / $0.502 → $0.132 / $0.5282026-08-20 23:36 UTC
- Hy3Baidubaidu/fp8$0.131 / $0.523 → $0.125 / $0.5022026-08-20 22:41 UTC
- Hy3Baidubaidu/fp8$0.132 / $0.528 → $0.131 / $0.5232026-08-20 21:41 UTC
- Hy3Baidubaidu/fp8$0.125 / $0.502 → $0.132 / $0.5282026-08-19 23:40 UTC
- Hy3Baidubaidu/fp8$0.131 / $0.523 → $0.125 / $0.5022026-08-19 21:40 UTC
- Hy3Baidubaidu/fp8$0.125 / $0.502 → $0.131 / $0.5232026-08-19 20:45 UTC
- Hy3Baidubaidu/fp8$0.131 / $0.523 → $0.125 / $0.5022026-08-19 19:40 UTC
- Hy3Baidubaidu/fp8$0.132 / $0.528 → $0.131 / $0.5232026-08-19 13:50 UTC
History
Availability
Axis starts at 90% so small differences stay visible; it expands to 0 when an endpoint falls below.
Time to first token
Logarithmic axis — the fleet spans two orders of magnitude. P50 is OpenRouter's rolling 30-minute window.
Throughput
Median tokens per second.