Qwen3 VL 235B A22B Instruct: 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) | 97.41% | 95.85% | 95.75% | 3.56s | 7.46s | 37 tps | 131K | $0.300 / $1.50 | |
| Degraded | 100%◆ (best) | 99.10% | 96.59% | 96.36% | 1.79s | 19.9s | 39 tps | 128K | $0.210 / $1.90 | |
| Healthy | 99.59% | 99.30% | 99.10%◆ (best) | 99.11% | 980ms | 7.31s | 44 tps | 131K | $0.210 / $1.90 | |
Alibabaalibaba | Healthy | 99.12% | 99.19% | 98.98% | 99.45%◆ (best) | 576ms◆ (best) | 2.39s◆ (best) | 47 tps◆ (best) | 131K | $0.260 / $1.04 |
| No data | — | 98.28% | 96.62% | 95.87% | 4.16s | 4.16s | 10 tps | 262K | $0.200 / $0.880◆ (best) | |
PrimeIntellectprimeintellect | No data | — | — | 98.03% | 98.03% | — | — | — | 262K | $0.300 / $1.50 |
- Novita100%◆ (best)Healthy24h 97.41%
- 7d
- 95.85%
- 30d
- 95.75%
- p50
- 3.56s
- p99
- 7.46s
- Throughput
- 37 tps
- Context
- 131K
- $/Mtok
- $0.300 / $1.50
novita/bf16bf16 - Venice100%◆ (best)Degraded24h 99.10%
- 7d
- 96.59%
- 30d
- 96.36%
- p50
- 1.79s
- p99
- 19.9s
- Throughput
- 39 tps
- Context
- 128K
- $/Mtok
- $0.210 / $1.90
venice/fp8fp8 - Parasail99.59%Healthy24h 99.30%
- 7d
- 99.10%◆ (best)
- 30d
- 99.11%
- p50
- 980ms
- p99
- 7.31s
- Throughput
- 44 tps
- Context
- 131K
- $/Mtok
- $0.210 / $1.90
parasail/fp8fp8 - Alibaba99.12%Healthy24h 99.19%
- 7d
- 98.98%
- 30d
- 99.45%◆ (best)
- p50
- 576ms◆ (best)
- p99
- 2.39s◆ (best)
- Throughput
- 47 tps◆ (best)
- Context
- 131K
- $/Mtok
- $0.260 / $1.04
alibaba - No data24h 98.28%
- 7d
- 96.62%
- 30d
- 95.87%
- p50
- 4.16s
- p99
- 4.16s
- Throughput
- 10 tps
- Context
- 262K
- $/Mtok
- $0.200 / $0.880◆ (best)
deepinfra/fp8fp8 - No data24h —
- 7d
- 98.03%
- 30d
- 98.03%
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 262K
- $/Mtok
- $0.300 / $1.50
primeintellect
Measured 2m 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, averaged over the hour.
Throughput
Median tokens per second.