Qwen3 Coder 480B A35B: 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 | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Healthy | 100%◆ (best) | 98.58% | 96.15% | 96.05% | 659ms◆ (best) | 16.6s | 72 tps | 262K | $0.300 / $1.00 | |
Googlegoogle-vertex/us-south1 | Healthy | 100%◆ (best) | 99.96% | 99.85% | 99.90%◆ (best) | 744ms | 3.86s◆ (best) | 67 tps | 262K | $0.220 / $1.80◆ (best) |
Alibabaalibaba/opensource | No data | — | 99.99% | 99.96%◆ (best) | 99.75% | 1.86s | 5.57s | 79 tps◆ (best) | 262K | $0.975 / $4.88 |
| No data | — | 94.68% | 64.26% | 84.91% | 4.18s | 38.3s | 22 tps | 262K | $0.380 / $1.55 | |
| No data | — | 96.84% | 90.23% | 89.48% | 1.30s | 4.52s | 50 tps | 256K | $0.350 / $1.50 |
- DeepInfra100%◆ (best)Healthy24h 98.58%
- 7d
- 96.15%
- 30d
- 96.05%
- p50
- 659ms◆ (best)
- p99
- 16.6s
- Throughput
- 72 tps
- Context
- 262K
- $/Mtok
- $0.300 / $1.00
deepinfra/turbofp4 - Google100%◆ (best)Healthy24h 99.96%
- 7d
- 99.85%
- 30d
- 99.90%◆ (best)
- p50
- 744ms
- p99
- 3.86s◆ (best)
- Throughput
- 67 tps
- Context
- 262K
- $/Mtok
- $0.220 / $1.80◆ (best)
google-vertex/us-south1 - No data24h 99.99%
- 7d
- 99.96%◆ (best)
- 30d
- 99.75%
- p50
- 1.86s
- p99
- 5.57s
- Throughput
- 79 tps◆ (best)
- Context
- 262K
- $/Mtok
- $0.975 / $4.88
alibaba/opensource - No data24h 94.68%
- 7d
- 64.26%
- 30d
- 84.91%
- p50
- 4.18s
- p99
- 38.3s
- Throughput
- 22 tps
- Context
- 262K
- $/Mtok
- $0.380 / $1.55
novita/fp8fp8 - No data24h 96.84%
- 7d
- 90.23%
- 30d
- 89.48%
- p50
- 1.30s
- p99
- 4.52s
- Throughput
- 50 tps
- Context
- 256K
- $/Mtok
- $0.350 / $1.50
venice/fp8fp8
Measured 4m 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.