Llama 3.1 8B 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) | 99.79% | 99.89% | 99.89% | 534ms | 3.24s | 16 tps | 32K | $0.152 / $0.287 | |
| Healthy | 100%◆ (best) | 100% | 100%◆ (best) | 100%◆ (best) | 176ms◆ (best) | 1.67s | 98 tps | 128K | $0.220 / $0.220 | |
| Healthy | 100%◆ (best) | 99.91% | 99.42% | 99.42% | 598ms | 18.5s | 17 tps | 131K | $0.020 / $0.040◆ (best) | |
Groqgroq | Healthy | 99.98% | 99.93% | 99.90% | 99.90% | 244ms | 1.35s◆ (best) | 114 tps◆ (best) | 131K | $0.050 / $0.080 |
| Healthy | 99.95% | 99.80% | 99.78% | 99.78% | 493ms | 2.72s | 52 tps | 16K | $0.020 / $0.050◆ (best) |
- Cloudflare100%◆ (best)Healthy24h 99.79%
- 7d
- 99.89%
- 30d
- 99.89%
- p50
- 534ms
- p99
- 3.24s
- Throughput
- 16 tps
- Context
- 32K
- $/Mtok
- $0.152 / $0.287
cloudflare/fp8fp8 - CoreWeave100%◆ (best)Healthy24h 100%
- 7d
- 100%◆ (best)
- 30d
- 100%◆ (best)
- p50
- 176ms◆ (best)
- p99
- 1.67s
- Throughput
- 98 tps
- Context
- 128K
- $/Mtok
- $0.220 / $0.220
coreweave/bf16bf16 - DeepInfra100%◆ (best)Healthy24h 99.91%
- 7d
- 99.42%
- 30d
- 99.42%
- p50
- 598ms
- p99
- 18.5s
- Throughput
- 17 tps
- Context
- 131K
- $/Mtok
- $0.020 / $0.040◆ (best)
deepinfra/fp8fp8 - Groq99.98%Healthy24h 99.93%
- 7d
- 99.90%
- 30d
- 99.90%
- p50
- 244ms
- p99
- 1.35s◆ (best)
- Throughput
- 114 tps◆ (best)
- Context
- 131K
- $/Mtok
- $0.050 / $0.080
groq - Novita99.95%Healthy24h 99.80%
- 7d
- 99.78%
- 30d
- 99.78%
- p50
- 493ms
- p99
- 2.72s
- Throughput
- 52 tps
- Context
- 16K
- $/Mtok
- $0.020 / $0.050◆ (best)
novita/fp8fp8
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.