Llama 3.3 70B Instruct
meta-llama/llama-3.3-70b-instruct
Compare with Llama 3.2 3B Instruct, Llama 3.2 1B Instruct, Llama 3.1 8B Instruct
CompareLlama 3.3 70B Instruct
| State | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Healthy | 100% | 95.89% | 94.19% | 94.19% | 661ms | 5.80s | 36 tps | 24K | $0.293 / $2.25 | |
| Healthy | 100% | 98.25% | 97.55% | 97.55% | 244ms | 905ms | 55 tps | 128K | $0.710 / $0.710 | |
| Healthy | 100% | 99.56% | 98.96% | 98.96% | 371ms | 1.89s | 31 tps | 131K | $0.250 / $0.750 | |
| Healthy | 100% | 97.87% | 97.79% | 97.79% | 569ms | 5.99s | 29 tps | 131K | $0.220 / $0.500 | |
Groqgroq | Healthy | 99.58% | 99.91% | 99.58% | 99.58% | 172ms | 1.08s | 103 tps | 131K | $0.590 / $0.790 |
| Degraded | 96.32% | 95.30% | 94.50% | 94.50% | 667ms | 3.92s | 37 tps | 12K | $0.135 / $0.400 | |
SambaNovasambanova-turbo | Healthy | 95.45% | 96.56% | 95.20% | 95.20% | 793ms | 5.52s | 49 tps | 131K | $0.450 / $0.900 |
| Degraded | 92.98% | 96.61% | 96.45% | 96.45% | 789ms | 10.7s | 11 tps | 131K | $0.100 / $0.320 | |
| Down | 89.49% | 97.38% | 97.60% | 97.60% | 1.53s | 9.65s | 16 tps | 131K | $0.200 / $0.520 | |
Googlegoogle-vertex | No data | — | 99.14% | 98.28% | 98.28% | 301ms | 1.66s | 49 tps | 128K | $0.720 / $0.720 |
Googlegoogle-vertex/us-central1 | No data | — | 100% | 100% | 100% | 301ms | 1.35s | 55 tps | 128K | $0.720 / $0.720 |
| Down | — | 41.71% | 42.99% | 42.99% | 7.44s | 40.5s | 0.0 tps | 131K | $0.130 / $0.400 | |
Togethertogether | No data | — | 97.93% | 97.66% | 97.66% | 761ms | 3.04s | 18 tps | 131K | $1.04 / $1.04 |
- Cloudflare100%Healthy24h 95.89%
- 7d
- 94.19%
- 30d
- 94.19%
- p50
- 661ms
- p99
- 5.80s
- Throughput
- 36 tps
- Context
- 24K
- $/Mtok
- $0.293 / $2.25
cloudflare/fp8fp8 - CoreWeave100%Healthy24h 98.25%
- 7d
- 97.55%
- 30d
- 97.55%
- p50
- 244ms
- p99
- 905ms
- Throughput
- 55 tps
- Context
- 128K
- $/Mtok
- $0.710 / $0.710
coreweave/fp16fp16 - Crusoe100%Healthy24h 99.56%
- 7d
- 98.96%
- 30d
- 98.96%
- p50
- 371ms
- p99
- 1.89s
- Throughput
- 31 tps
- Context
- 131K
- $/Mtok
- $0.250 / $0.750
crusoe/bf16bf16 - Parasail100%Healthy24h 97.87%
- 7d
- 97.79%
- 30d
- 97.79%
- p50
- 569ms
- p99
- 5.99s
- Throughput
- 29 tps
- Context
- 131K
- $/Mtok
- $0.220 / $0.500
parasail/fp8fp8 - Groq99.58%Healthy24h 99.91%
- 7d
- 99.58%
- 30d
- 99.58%
- p50
- 172ms
- p99
- 1.08s
- Throughput
- 103 tps
- Context
- 131K
- $/Mtok
- $0.590 / $0.790
groq - Novita96.32%Degraded24h 95.30%
- 7d
- 94.50%
- 30d
- 94.50%
- p50
- 667ms
- p99
- 3.92s
- Throughput
- 37 tps
- Context
- 12K
- $/Mtok
- $0.135 / $0.400
novita/bf16bf16 - SambaNova95.45%Healthy24h 96.56%
- 7d
- 95.20%
- 30d
- 95.20%
- p50
- 793ms
- p99
- 5.52s
- Throughput
- 49 tps
- Context
- 131K
- $/Mtok
- $0.450 / $0.900
sambanova-turbo - DeepInfra92.98%Degraded24h 96.61%
- 7d
- 96.45%
- 30d
- 96.45%
- p50
- 789ms
- p99
- 10.7s
- Throughput
- 11 tps
- Context
- 131K
- $/Mtok
- $0.100 / $0.320
deepinfra/turbofp8 - AkashML89.49%Down24h 97.38%
- 7d
- 97.60%
- 30d
- 97.60%
- p50
- 1.53s
- p99
- 9.65s
- Throughput
- 16 tps
- Context
- 131K
- $/Mtok
- $0.200 / $0.520
akashml/fp8fp8 - No data24h 99.14%
- 7d
- 98.28%
- 30d
- 98.28%
- p50
- 301ms
- p99
- 1.66s
- Throughput
- 49 tps
- Context
- 128K
- $/Mtok
- $0.720 / $0.720
google-vertex - No data24h 100%
- 7d
- 100%
- 30d
- 100%
- p50
- 301ms
- p99
- 1.35s
- Throughput
- 55 tps
- Context
- 128K
- $/Mtok
- $0.720 / $0.720
google-vertex/us-central1 - Down24h 41.71%
- 7d
- 42.99%
- 30d
- 42.99%
- p50
- 7.44s
- p99
- 40.5s
- Throughput
- 0.0 tps
- Context
- 131K
- $/Mtok
- $0.130 / $0.400
nebius/fp8fp8 - No data24h 97.93%
- 7d
- 97.66%
- 30d
- 97.66%
- p50
- 761ms
- p99
- 3.04s
- Throughput
- 18 tps
- Context
- 131K
- $/Mtok
- $1.04 / $1.04
together
Measured 1m ago. Latency and throughput are OpenRouter's rolling 30-minute windows.
Price changes
- Llama 3.3 70B InstructAkashMLakashml/fp8$0.130 / $0.400 → $0.200 / $0.5202026-08-19 09:50 UTC
History
3 endpoints reporting no data are omitted from the charts.
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.