gpt-oss-20b: 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 | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
Amazon Bedrockamazon-bedrock | Healthy | 100%◆ (best) | 99.98% | 99.97% | 99.37% | 558ms | 4.38s | 256 tps◆ (best) | 131K | $0.070 / $0.150 |
| Healthy | 100%◆ (best) | 99.95% | 99.99%◆ (best) | 99.88% | 118ms◆ (best) | 4.49s | 154 tps | 131K | $0.030 / $0.130 | |
| Healthy | 100%◆ (best) | 99.93% | 99.88% | 99.89%◆ (best) | 258ms | 4.68s | 96 tps | 131K | $0.030 / $0.140 | |
| Healthy | 100%◆ (best) | 99.89% | 99.77% | 99.76% | 675ms | 5.93s | 149 tps | 131K | $0.040 / $0.150 | |
| Healthy | 99.91% | 99.87% | 99.73% | 99.24% | 1.11s | 6.16s | 81 tps | 131K | $0.030 / $0.150 | |
| Healthy | 99.81% | 99.35% | 99.43% | 99.51% | 844ms | 29.2s | 22 tps | 131K | $0.029 / $0.140 | |
| Healthy | 99.74% | 99.11% | 99.29% | 99.00% | 818ms | 24.3s | 32 tps | 131K | $0.020 / $0.100 | |
| Healthy | 98.65% | 99.62% | 99.80% | 99.67% | 2.72s | 26.5s | 24 tps | 131K | $0.018 / $0.090◆ (best) | |
Groqgroq | Healthy | 98.08% | 97.73% | 96.55% | 98.50% | 886ms | 4.19s◆ (best) | 235 tps | 131K | $0.075 / $0.300 |
| Down | 46.98% | 85.46% | 93.09% | 95.46% | 1.12s | 12.6s | 32 tps | 131K | $0.040 / $0.180 | |
| No data | — | — | — | — | — | — | — | 131K | $0.020 / $0.100 | |
Amazon Bedrockamazon-bedrock/eu-west-1 | No data | — | 100% | 99.72% | 99.62% | — | — | — | 131K | $0.070 / $0.150 |
Fireworksfireworks | No data | — | — | — | 0.00% | — | — | — | 131K | $0.070 / $0.300 |
Googlegoogle-vertex/us-central1 | Down | — | 95.99% | 93.78% | 96.55% | 291ms | 8.69s | 200 tps | 131K | $0.070 / $0.250 |
Phalaphala | No data | — | — | — | 99.44% | — | — | — | 131K | $0.040 / $0.150 |
PrimeIntellectprimeintellect | No data | — | — | 99.61% | 99.61% | — | — | — | 131K | $0.070 / $0.300 |
Togethertogether | No data | — | — | — | 0.00% | — | — | — | 131K | $0.050 / $0.200 |
- Amazon Bedrock100%◆ (best)Healthy24h 99.98%
- 7d
- 99.97%
- 30d
- 99.37%
- p50
- 558ms
- p99
- 4.38s
- Throughput
- 256 tps◆ (best)
- Context
- 131K
- $/Mtok
- $0.070 / $0.150
amazon-bedrock - CoreWeave100%◆ (best)Healthy24h 99.95%
- 7d
- 99.99%◆ (best)
- 30d
- 99.88%
- p50
- 118ms◆ (best)
- p99
- 4.49s
- Throughput
- 154 tps
- Context
- 131K
- $/Mtok
- $0.030 / $0.130
coreweave/fp4fp4 - DeepInfra100%◆ (best)Healthy24h 99.93%
- 7d
- 99.88%
- 30d
- 99.89%◆ (best)
- p50
- 258ms
- p99
- 4.68s
- Throughput
- 96 tps
- Context
- 131K
- $/Mtok
- $0.030 / $0.140
deepinfra/bf16bf16 - Novita100%◆ (best)Healthy24h 99.89%
- 7d
- 99.77%
- 30d
- 99.76%
- p50
- 675ms
- p99
- 5.93s
- Throughput
- 149 tps
- Context
- 131K
- $/Mtok
- $0.040 / $0.150
novita/fp4fp4 - Parasail99.91%Healthy24h 99.87%
- 7d
- 99.73%
- 30d
- 99.24%
- p50
- 1.11s
- p99
- 6.16s
- Throughput
- 81 tps
- Context
- 131K
- $/Mtok
- $0.030 / $0.150
parasail/fp4fp4 - DekaLLM99.81%Healthy24h 99.35%
- 7d
- 99.43%
- 30d
- 99.51%
- p50
- 844ms
- p99
- 29.2s
- Throughput
- 22 tps
- Context
- 131K
- $/Mtok
- $0.029 / $0.140
dekallm/bf16bf16 - AkashML99.74%Healthy24h 99.11%
- 7d
- 99.29%
- 30d
- 99.00%
- p50
- 818ms
- p99
- 24.3s
- Throughput
- 32 tps
- Context
- 131K
- $/Mtok
- $0.020 / $0.100
akashml/fp4fp4 - Darkbloom98.65%Healthy24h 99.62%
- 7d
- 99.80%
- 30d
- 99.67%
- p50
- 2.72s
- p99
- 26.5s
- Throughput
- 24 tps
- Context
- 131K
- $/Mtok
- $0.018 / $0.090◆ (best)
darkbloom/fp8fp8 - Groq98.08%Healthy24h 97.73%
- 7d
- 96.55%
- 30d
- 98.50%
- p50
- 886ms
- p99
- 4.19s◆ (best)
- Throughput
- 235 tps
- Context
- 131K
- $/Mtok
- $0.075 / $0.300
groq - SiliconFlow46.98%Down24h 85.46%
- 7d
- 93.09%
- 30d
- 95.46%
- p50
- 1.12s
- p99
- 12.6s
- Throughput
- 32 tps
- Context
- 131K
- $/Mtok
- $0.040 / $0.180
siliconflow/fp8fp8 - No data24h —
- 7d
- —
- 30d
- —
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 131K
- $/Mtok
- $0.020 / $0.100
akashml/bf16bf16 - No data24h 100%
- 7d
- 99.72%
- 30d
- 99.62%
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 131K
- $/Mtok
- $0.070 / $0.150
amazon-bedrock/eu-west-1 - No data24h —
- 7d
- —
- 30d
- 0.00%
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 131K
- $/Mtok
- $0.070 / $0.300
fireworks - Down24h 95.99%
- 7d
- 93.78%
- 30d
- 96.55%
- p50
- 291ms
- p99
- 8.69s
- Throughput
- 200 tps
- Context
- 131K
- $/Mtok
- $0.070 / $0.250
google-vertex/us-central1 - No data24h —
- 7d
- —
- 30d
- 99.44%
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 131K
- $/Mtok
- $0.040 / $0.150
phala - No data24h —
- 7d
- 99.61%
- 30d
- 99.61%
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 131K
- $/Mtok
- $0.070 / $0.300
primeintellect - No data24h —
- 7d
- —
- 30d
- 0.00%
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 131K
- $/Mtok
- $0.050 / $0.200
together
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, averaged over the hour.
Throughput
Median tokens per second.