DeepSeek V4.1 Flash 1017: 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 | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
BaseTenbaseten | Healthy | 100%◆ (best) | 100% | 100%◆ (best) | 100%◆ (best) | 312ms◆ (best) | 5.60s | 312 tps◆ (best) | 1.0M | $0.300 / $1.20 |
Cloudflarecloudflare | Healthy | 100%◆ (best) | 98.52% | 100%◆ (best) | 100%◆ (best) | 2.22s | 32.1s | 84 tps | 1.0M | $0.100 / $0.600 |
| Healthy | 100%◆ (best) | 98.54% | 100%◆ (best) | 100%◆ (best) | 695ms | 4.22s | 99 tps | 1.0M | $0.098 / $0.294 | |
| Healthy | 100%◆ (best) | 95.29% | 100%◆ (best) | 100%◆ (best) | 4.42s | 18.0s | 64 tps | 1.0M | $0.078 / $0.235 | |
| Healthy | 100%◆ (best) | 98.05% | 100%◆ (best) | 100%◆ (best) | 1.06s | 6.18s | 138 tps | 1.0M | $0.147 / $0.441 | |
DeepSeekdeepseek | Healthy | 100%◆ (best) | 100% | 100%◆ (best) | 100%◆ (best) | 1.00s | 2.98s◆ (best) | 50 tps | 1.0M | $0.150 / $0.600 |
Fireworksfireworks | Healthy | 100%◆ (best) | 100% | 100%◆ (best) | 100%◆ (best) | 585ms | 11.0s | 141 tps | 1.0M | $0.300 / $1.20 |
| Healthy | 100%◆ (best) | 100% | 100%◆ (best) | 100%◆ (best) | 435ms | 4.45s | 168 tps | 262K | $0.080 / $0.500 | |
| Healthy | 100%◆ (best) | 100% | 100%◆ (best) | 100%◆ (best) | 880ms | 15.8s | 137 tps | 1.0M | $0.300 / $1.20 | |
PrimeIntellectprimeintellect | Healthy | 100%◆ (best) | 99.33% | 99.05% | 99.05% | 1.30s | 20.8s | 124 tps | 1.0M | $0.300 / $1.20 |
Relacerelace | Healthy | 100%◆ (best) | 100% | 100%◆ (best) | 100%◆ (best) | 1.02s | 5.88s | 130 tps | 1.0M | $0.050 / $0.600◆ (best) |
| No data | — | — | 100%◆ (best) | 100%◆ (best) | — | — | — | 1.0M | $0.081 / $0.550 |
- BaseTen100%◆ (best)Healthy24h 100%
- 7d
- 100%◆ (best)
- 30d
- 100%◆ (best)
- p50
- 312ms◆ (best)
- p99
- 5.60s
- Throughput
- 312 tps◆ (best)
- Context
- 1.0M
- $/Mtok
- $0.300 / $1.20
baseten - Cloudflare100%◆ (best)Healthy24h 98.52%
- 7d
- 100%◆ (best)
- 30d
- 100%◆ (best)
- p50
- 2.22s
- p99
- 32.1s
- Throughput
- 84 tps
- Context
- 1.0M
- $/Mtok
- $0.100 / $0.600
cloudflare - DeepInfra100%◆ (best)Healthy24h 98.54%
- 7d
- 100%◆ (best)
- 30d
- 100%◆ (best)
- p50
- 695ms
- p99
- 4.22s
- Throughput
- 99 tps
- Context
- 1.0M
- $/Mtok
- $0.098 / $0.294
deepinfra/fp8fp8 - DeepInfra100%◆ (best)Healthy24h 95.29%
- 7d
- 100%◆ (best)
- 30d
- 100%◆ (best)
- p50
- 4.42s
- p99
- 18.0s
- Throughput
- 64 tps
- Context
- 1.0M
- $/Mtok
- $0.078 / $0.235
deepinfra/fp8/flexfp8 - DeepInfra100%◆ (best)Healthy24h 98.05%
- 7d
- 100%◆ (best)
- 30d
- 100%◆ (best)
- p50
- 1.06s
- p99
- 6.18s
- Throughput
- 138 tps
- Context
- 1.0M
- $/Mtok
- $0.147 / $0.441
deepinfra/fp8/priorityfp8 - DeepSeek100%◆ (best)Healthy24h 100%
- 7d
- 100%◆ (best)
- 30d
- 100%◆ (best)
- p50
- 1.00s
- p99
- 2.98s◆ (best)
- Throughput
- 50 tps
- Context
- 1.0M
- $/Mtok
- $0.150 / $0.600
deepseek - Fireworks100%◆ (best)Healthy24h 100%
- 7d
- 100%◆ (best)
- 30d
- 100%◆ (best)
- p50
- 585ms
- p99
- 11.0s
- Throughput
- 141 tps
- Context
- 1.0M
- $/Mtok
- $0.300 / $1.20
fireworks - InferenceNet100%◆ (best)Healthy24h 100%
- 7d
- 100%◆ (best)
- 30d
- 100%◆ (best)
- p50
- 435ms
- p99
- 4.45s
- Throughput
- 168 tps
- Context
- 262K
- $/Mtok
- $0.080 / $0.500
inference-net/fp8fp8 - Parasail100%◆ (best)Healthy24h 100%
- 7d
- 100%◆ (best)
- 30d
- 100%◆ (best)
- p50
- 880ms
- p99
- 15.8s
- Throughput
- 137 tps
- Context
- 1.0M
- $/Mtok
- $0.300 / $1.20
parasail/fp8fp8 - PrimeIntellect100%◆ (best)Healthy24h 99.33%
- 7d
- 99.05%
- 30d
- 99.05%
- p50
- 1.30s
- p99
- 20.8s
- Throughput
- 124 tps
- Context
- 1.0M
- $/Mtok
- $0.300 / $1.20
primeintellect - Relace100%◆ (best)Healthy24h 100%
- 7d
- 100%◆ (best)
- 30d
- 100%◆ (best)
- p50
- 1.02s
- p99
- 5.88s
- Throughput
- 130 tps
- Context
- 1.0M
- $/Mtok
- $0.050 / $0.600◆ (best)
relace - No data24h —
- 7d
- 100%◆ (best)
- 30d
- 100%◆ (best)
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 1.0M
- $/Mtok
- $0.081 / $0.550
morph/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.