← Qwen3 235B A22B Instruct 2507
Qwen3 235B A22B Instruct 2507: providers compared
Every endpoint serving this model, side by side. ◆ marks the best value in each column — per model, never fleet-wide.
Current state
| State | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
Alibabaalibaba | Healthy | 100%◆ (best) | 99.98% | 99.98%◆ (best) | 99.98%◆ (best) | 490ms | 2.01s◆ (best) | 43 tps◆ (best) | 131K | $0.150 / $0.598 |
| Healthy | 100%◆ (best) | 99.53% | 99.81% | 99.81% | 1.02s | 5.92s | 27 tps | 131K | $0.200 / $0.880 | |
| Down | 100%◆ (best) | 96.46% | 96.32% | 96.32% | 416ms◆ (best) | 18.8s | 14 tps | 262K | $0.090 / $0.550◆ (best) | |
Googlegoogle-vertex/us-south1 | Healthy | 100%◆ (best) | 99.15% | 98.76% | 98.76% | 534ms | 15.6s | 38 tps | 262K | $0.220 / $0.880 |
| Degraded | 100%◆ (best) | 96.40% | 95.22% | 95.22% | 604ms | 28.6s | 31 tps | 262K | $0.200 / $0.600 | |
| Degraded | 100%◆ (best) | 97.95% | 98.73% | 98.73% | 680ms | 11.7s | 29 tps | 131K | $0.090 / $0.580◆ (best) | |
| Healthy | 100%◆ (best) | 99.61% | 99.23% | 99.23% | 515ms | 3.35s | 26 tps | 131K | $0.140 / $0.800 | |
StreamLakestreamlake | Healthy | 100%◆ (best) | 99.79% | 99.83% | 99.83% | 966ms | 3.72s | 25 tps | 128K | $0.210 / $0.840 |
| Healthy | 99.76% | 99.69% | 99.72% | 99.72% | 1.57s | 4.66s | 25 tps | 262K | $0.350 / $1.40 | |
| Down | — | 95.67% | 95.64% | 95.64% | 4.70s | 35.8s | 3.0 tps | 128K | $0.150 / $0.750 |
- Alibaba100%◆ (best)Healthy24h 99.98%
- 7d
- 99.98%◆ (best)
- 30d
- 99.98%◆ (best)
- p50
- 490ms
- p99
- 2.01s◆ (best)
- Throughput
- 43 tps◆ (best)
- Context
- 131K
- $/Mtok
- $0.150 / $0.598
alibaba - AtlasCloud100%◆ (best)Healthy24h 99.53%
- 7d
- 99.81%
- 30d
- 99.81%
- p50
- 1.02s
- p99
- 5.92s
- Throughput
- 27 tps
- Context
- 131K
- $/Mtok
- $0.200 / $0.880
atlas-cloud/fp8fp8 - DeepInfra100%◆ (best)Down24h 96.46%
- 7d
- 96.32%
- 30d
- 96.32%
- p50
- 416ms◆ (best)
- p99
- 18.8s
- Throughput
- 14 tps
- Context
- 262K
- $/Mtok
- $0.090 / $0.550◆ (best)
deepinfra/fp8fp8 - Google100%◆ (best)Healthy24h 99.15%
- 7d
- 98.76%
- 30d
- 98.76%
- p50
- 534ms
- p99
- 15.6s
- Throughput
- 38 tps
- Context
- 262K
- $/Mtok
- $0.220 / $0.880
google-vertex/us-south1 - Nebius100%◆ (best)Degraded24h 96.40%
- 7d
- 95.22%
- 30d
- 95.22%
- p50
- 604ms
- p99
- 28.6s
- Throughput
- 31 tps
- Context
- 262K
- $/Mtok
- $0.200 / $0.600
nebius/fp8fp8 - Novita100%◆ (best)Degraded24h 97.95%
- 7d
- 98.73%
- 30d
- 98.73%
- p50
- 680ms
- p99
- 11.7s
- Throughput
- 29 tps
- Context
- 131K
- $/Mtok
- $0.090 / $0.580◆ (best)
novita/fp8fp8 - Parasail100%◆ (best)Healthy24h 99.61%
- 7d
- 99.23%
- 30d
- 99.23%
- p50
- 515ms
- p99
- 3.35s
- Throughput
- 26 tps
- Context
- 131K
- $/Mtok
- $0.140 / $0.800
parasail/fp8fp8 - StreamLake100%◆ (best)Healthy24h 99.79%
- 7d
- 99.83%
- 30d
- 99.83%
- p50
- 966ms
- p99
- 3.72s
- Throughput
- 25 tps
- Context
- 128K
- $/Mtok
- $0.210 / $0.840
streamlake - GMICloud99.76%Healthy24h 99.69%
- 7d
- 99.72%
- 30d
- 99.72%
- p50
- 1.57s
- p99
- 4.66s
- Throughput
- 25 tps
- Context
- 262K
- $/Mtok
- $0.350 / $1.40
gmicloud/fp8fp8 - Down24h 95.67%
- 7d
- 95.64%
- 30d
- 95.64%
- p50
- 4.70s
- p99
- 35.8s
- Throughput
- 3.0 tps
- Context
- 128K
- $/Mtok
- $0.150 / $0.750
venice/fp8fp8
Measured 1m ago. Latency and throughput are OpenRouter's rolling 30-minute windows. ◆ marks the best value in each column among these endpoints.
Price vs latency
One dot per endpoint at the latest round · further left = faster, lower = cheaper · hover a dot for its tag
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.