gpt-oss-120b: 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) | 99.67% | 99.18% | 99.18% | 258ms | 1.09s◆ (best) | 199 tps | 128K | $0.100 / $0.500 | |
| Healthy | 100%◆ (best) | 99.99% | 99.99% | 99.99% | 242ms◆ (best) | 2.38s | 908 tps◆ (best) | 131K | $0.350 / $0.750 | |
| Healthy | 100%◆ (best) | 95.88% | 97.79% | 97.79% | 343ms | 1.53s | 45 tps | 131K | $0.037 / $0.170 | |
| Healthy | 100%◆ (best) | 99.70% | 99.01% | 99.01% | 432ms | 1.69s | 104 tps | 131K | $0.150 / $0.600 | |
Phalaphala | Healthy | 100%◆ (best) | 97.42% | 97.66% | 97.66% | 917ms | 6.47s | 69 tps | 131K | $0.150 / $0.600 |
| Healthy | 99.98% | 99.71% | 99.84% | 99.84% | 324ms | 1.79s | 90 tps | 131K | $0.100 / $0.750 | |
DigitalOceandigitalocean | Healthy | 99.97% | 99.90% | 99.90% | 99.90% | 940ms | 4.59s | 19 tps | 128K | $0.055 / $0.385 |
| Healthy | 99.96% | 99.19% | 99.59% | 99.59% | 1.32s | 4.18s | 57 tps | 131K | $0.037 / $0.490 | |
Groqgroq | Healthy | 99.96% | 99.94% | 99.93% | 99.93% | 267ms | 2.49s | 328 tps | 131K | $0.150 / $0.600 |
| Healthy | 99.77% | 99.26% | 99.75% | 99.75% | 892ms | 4.95s | 24 tps | 131K | $0.085 / $0.500 | |
Amazon Bedrockamazon-bedrock | Healthy | 99.72% | 95.19% | 99.02% | 99.02% | 405ms | 1.71s | 194 tps | 131K | $0.150 / $0.600 |
| Healthy | 99.65% | 98.97% | 98.80% | 98.80% | 390ms | 9.57s | 143 tps | 131K | $0.150 / $0.600 | |
| Healthy | 99.50% | 95.58% | 96.24% | 96.24% | 486ms | 7.00s | 28 tps | 131K | $0.030 / $0.170◆ (best) | |
SambaNovasambanova | Degraded | 96.73% | 88.75% | 86.10% | 86.10% | 967ms | 8.51s | 233 tps | 131K | $0.140 / $0.950 |
Maramara | Healthy | 96.41% | 97.24% | 95.51% | 95.51% | 3.88s | 13.3s | 88 tps | 131K | $0.150 / $0.750 |
Togethertogether | Degraded | 90.77% | 95.31% | 97.39% | 97.39% | 371ms | 4.94s | 46 tps | 131K | $0.150 / $0.600 |
Googlegoogle-vertex/global | Down | 85.12% | 89.80% | 89.93% | 89.93% | 1.40s | 10.6s | 45 tps | 131K | $0.090 / $0.360 |
| Down | 61.13% | 87.73% | 93.08% | 93.08% | 585ms | 13.8s | 42 tps | 131K | $0.050 / $0.250 | |
| Down | 2.01% | 78.20% | 92.45% | 92.45% | 2.14s | 23.7s | 11 tps | 131K | $0.050 / $0.450 | |
Amazon Bedrockamazon-bedrock/eu-west-1 | No data | — | — | 100%◆ (best) | 100%◆ (best) | — | — | — | 131K | $0.150 / $0.600 |
- BaseTen100%◆ (best)Healthy24h 99.67%
- 7d
- 99.18%
- 30d
- 99.18%
- p50
- 258ms
- p99
- 1.09s◆ (best)
- Throughput
- 199 tps
- Context
- 128K
- $/Mtok
- $0.100 / $0.500
baseten/fp4fp4 - Cerebras100%◆ (best)Healthy24h 99.99%
- 7d
- 99.99%
- 30d
- 99.99%
- p50
- 242ms◆ (best)
- p99
- 2.38s
- Throughput
- 908 tps◆ (best)
- Context
- 131K
- $/Mtok
- $0.350 / $0.750
cerebras/fp16fp16 - DeepInfra100%◆ (best)Healthy24h 95.88%
- 7d
- 97.79%
- 30d
- 97.79%
- p50
- 343ms
- p99
- 1.53s
- Throughput
- 45 tps
- Context
- 131K
- $/Mtok
- $0.037 / $0.170
deepinfra/bf16bf16 - DeepInfra100%◆ (best)Healthy24h 99.70%
- 7d
- 99.01%
- 30d
- 99.01%
- p50
- 432ms
- p99
- 1.69s
- Throughput
- 104 tps
- Context
- 131K
- $/Mtok
- $0.150 / $0.600
deepinfra/turbobf16 - Phala100%◆ (best)Healthy24h 97.42%
- 7d
- 97.66%
- 30d
- 97.66%
- p50
- 917ms
- p99
- 6.47s
- Throughput
- 69 tps
- Context
- 131K
- $/Mtok
- $0.150 / $0.600
phala - Parasail99.98%Healthy24h 99.71%
- 7d
- 99.84%
- 30d
- 99.84%
- p50
- 324ms
- p99
- 1.79s
- Throughput
- 90 tps
- Context
- 131K
- $/Mtok
- $0.100 / $0.750
parasail/fp4fp4 - DigitalOcean99.97%Healthy24h 99.90%
- 7d
- 99.90%
- 30d
- 99.90%
- p50
- 940ms
- p99
- 4.59s
- Throughput
- 19 tps
- Context
- 128K
- $/Mtok
- $0.055 / $0.385
digitalocean - AkashML99.96%Healthy24h 99.19%
- 7d
- 99.59%
- 30d
- 99.59%
- p50
- 1.32s
- p99
- 4.18s
- Throughput
- 57 tps
- Context
- 131K
- $/Mtok
- $0.037 / $0.490
akashml/bf16bf16 - Groq99.96%Healthy24h 99.94%
- 7d
- 99.93%
- 30d
- 99.93%
- p50
- 267ms
- p99
- 2.49s
- Throughput
- 328 tps
- Context
- 131K
- $/Mtok
- $0.150 / $0.600
groq - Mancer 299.77%Healthy24h 99.26%
- 7d
- 99.75%
- 30d
- 99.75%
- p50
- 892ms
- p99
- 4.95s
- Throughput
- 24 tps
- Context
- 131K
- $/Mtok
- $0.085 / $0.500
mancer/fp8fp8 - Amazon Bedrock99.72%Healthy24h 95.19%
- 7d
- 99.02%
- 30d
- 99.02%
- p50
- 405ms
- p99
- 1.71s
- Throughput
- 194 tps
- Context
- 131K
- $/Mtok
- $0.150 / $0.600
amazon-bedrock - Nebius99.65%Healthy24h 98.97%
- 7d
- 98.80%
- 30d
- 98.80%
- p50
- 390ms
- p99
- 9.57s
- Throughput
- 143 tps
- Context
- 131K
- $/Mtok
- $0.150 / $0.600
nebius/fp4fp4 - CoreWeave99.50%Healthy24h 95.58%
- 7d
- 96.24%
- 30d
- 96.24%
- p50
- 486ms
- p99
- 7.00s
- Throughput
- 28 tps
- Context
- 131K
- $/Mtok
- $0.030 / $0.170◆ (best)
coreweave/fp4fp4 - SambaNova96.73%Degraded24h 88.75%
- 7d
- 86.10%
- 30d
- 86.10%
- p50
- 967ms
- p99
- 8.51s
- Throughput
- 233 tps
- Context
- 131K
- $/Mtok
- $0.140 / $0.950
sambanova - Mara96.41%Healthy24h 97.24%
- 7d
- 95.51%
- 30d
- 95.51%
- p50
- 3.88s
- p99
- 13.3s
- Throughput
- 88 tps
- Context
- 131K
- $/Mtok
- $0.150 / $0.750
mara - Together90.77%Degraded24h 95.31%
- 7d
- 97.39%
- 30d
- 97.39%
- p50
- 371ms
- p99
- 4.94s
- Throughput
- 46 tps
- Context
- 131K
- $/Mtok
- $0.150 / $0.600
together - Google85.12%Down24h 89.80%
- 7d
- 89.93%
- 30d
- 89.93%
- p50
- 1.40s
- p99
- 10.6s
- Throughput
- 45 tps
- Context
- 131K
- $/Mtok
- $0.090 / $0.360
google-vertex/global - Novita61.13%Down24h 87.73%
- 7d
- 93.08%
- 30d
- 93.08%
- p50
- 585ms
- p99
- 13.8s
- Throughput
- 42 tps
- Context
- 131K
- $/Mtok
- $0.050 / $0.250
novita/fp4fp4 - SiliconFlow2.01%Down24h 78.20%
- 7d
- 92.45%
- 30d
- 92.45%
- p50
- 2.14s
- p99
- 23.7s
- Throughput
- 11 tps
- Context
- 131K
- $/Mtok
- $0.050 / $0.450
siliconflow/fp8fp8 - No data24h —
- 7d
- 100%◆ (best)
- 30d
- 100%◆ (best)
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 131K
- $/Mtok
- $0.150 / $0.600
amazon-bedrock/eu-west-1
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.