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% | 99.71% | 99.00% | 99.00% | 435ms | 1.80s | 28 tps | 131K | $0.250 / $0.750 | |
Googlegoogle-vertex/us-central1 | Healthy | 100% | 99.99% | 100% | 100% | 312ms | 2.00s | 49 tps | 128K | $0.720 / $0.720 |
Groqgroq | Healthy | 99.73% | 99.92% | 99.58% | 99.58% | 239ms | 1.76s | 150 tps | 131K | $0.590 / $0.790 |
| Healthy | 99.71% | 98.29% | 97.64% | 97.64% | 244ms | 898ms | 57 tps | 128K | $0.710 / $0.710 | |
| Degraded | 98.49% | 96.96% | 97.35% | 97.35% | 1.44s | 8.90s | 15 tps | 131K | $0.200 / $0.520 | |
| Healthy | 98.10% | 98.12% | 97.85% | 97.85% | 698ms | 6.84s | 33 tps | 131K | $0.220 / $0.500 | |
| Healthy | 98.02% | 95.86% | 94.29% | 94.29% | 673ms | 5.02s | 39 tps | 24K | $0.293 / $2.25 | |
| Healthy | 95.63% | 96.72% | 96.45% | 96.45% | 754ms | 6.97s | 13 tps | 131K | $0.100 / $0.320 | |
SambaNovasambanova-turbo | Healthy | 95.19% | 96.63% | 95.22% | 95.22% | 763ms | 4.38s | 57 tps | 131K | $0.450 / $0.900 |
| Down | 84.78% | 95.18% | 94.51% | 94.51% | 746ms | 10.5s | 34 tps | 12K | $0.135 / $0.400 | |
Googlegoogle-vertex | No data | — | 99.15% | 98.34% | 98.34% | 310ms | 1.36s | 51 tps | 128K | $0.720 / $0.720 |
| Down | — | 41.52% | 43.62% | 43.62% | 6.49s | 39.0s | 0.0 tps | 131K | $0.130 / $0.400 | |
Togethertogether | No data | — | 98.19% | 97.67% | 97.67% | 740ms | 5.02s | 42 tps | 131K | $1.04 / $1.04 |
- Crusoe100%Healthy24h 99.71%
- 7d
- 99.00%
- 30d
- 99.00%
- p50
- 435ms
- p99
- 1.80s
- Throughput
- 28 tps
- Context
- 131K
- $/Mtok
- $0.250 / $0.750
crusoe/bf16bf16 - Google100%Healthy24h 99.99%
- 7d
- 100%
- 30d
- 100%
- p50
- 312ms
- p99
- 2.00s
- Throughput
- 49 tps
- Context
- 128K
- $/Mtok
- $0.720 / $0.720
google-vertex/us-central1 - Groq99.73%Healthy24h 99.92%
- 7d
- 99.58%
- 30d
- 99.58%
- p50
- 239ms
- p99
- 1.76s
- Throughput
- 150 tps
- Context
- 131K
- $/Mtok
- $0.590 / $0.790
groq - CoreWeave99.71%Healthy24h 98.29%
- 7d
- 97.64%
- 30d
- 97.64%
- p50
- 244ms
- p99
- 898ms
- Throughput
- 57 tps
- Context
- 128K
- $/Mtok
- $0.710 / $0.710
coreweave/fp16fp16 - AkashML98.49%Degraded24h 96.96%
- 7d
- 97.35%
- 30d
- 97.35%
- p50
- 1.44s
- p99
- 8.90s
- Throughput
- 15 tps
- Context
- 131K
- $/Mtok
- $0.200 / $0.520
akashml/fp8fp8 - Parasail98.10%Healthy24h 98.12%
- 7d
- 97.85%
- 30d
- 97.85%
- p50
- 698ms
- p99
- 6.84s
- Throughput
- 33 tps
- Context
- 131K
- $/Mtok
- $0.220 / $0.500
parasail/fp8fp8 - Cloudflare98.02%Healthy24h 95.86%
- 7d
- 94.29%
- 30d
- 94.29%
- p50
- 673ms
- p99
- 5.02s
- Throughput
- 39 tps
- Context
- 24K
- $/Mtok
- $0.293 / $2.25
cloudflare/fp8fp8 - DeepInfra95.63%Healthy24h 96.72%
- 7d
- 96.45%
- 30d
- 96.45%
- p50
- 754ms
- p99
- 6.97s
- Throughput
- 13 tps
- Context
- 131K
- $/Mtok
- $0.100 / $0.320
deepinfra/turbofp8 - SambaNova95.19%Healthy24h 96.63%
- 7d
- 95.22%
- 30d
- 95.22%
- p50
- 763ms
- p99
- 4.38s
- Throughput
- 57 tps
- Context
- 131K
- $/Mtok
- $0.450 / $0.900
sambanova-turbo - Novita84.78%Down24h 95.18%
- 7d
- 94.51%
- 30d
- 94.51%
- p50
- 746ms
- p99
- 10.5s
- Throughput
- 34 tps
- Context
- 12K
- $/Mtok
- $0.135 / $0.400
novita/bf16bf16 - No data24h 99.15%
- 7d
- 98.34%
- 30d
- 98.34%
- p50
- 310ms
- p99
- 1.36s
- Throughput
- 51 tps
- Context
- 128K
- $/Mtok
- $0.720 / $0.720
google-vertex - Down24h 41.52%
- 7d
- 43.62%
- 30d
- 43.62%
- p50
- 6.49s
- p99
- 39.0s
- Throughput
- 0.0 tps
- Context
- 131K
- $/Mtok
- $0.130 / $0.400
nebius/fp8fp8 - No data24h 98.19%
- 7d
- 97.67%
- 30d
- 97.67%
- p50
- 740ms
- p99
- 5.02s
- Throughput
- 42 tps
- Context
- 131K
- $/Mtok
- $1.04 / $1.04
together
Measured 3m 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
2 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.