Llama 3.3 70B Instruct
meta-llama/llama-3.3-70b-instructInputOutput
Compare its 14 providersIncident RSSCompare with Llama 3.2 3B Instruct, Llama 3.2 1B Instruct, Llama 3.1 8B Instruct
CompareLlama 3.3 70B Instruct
| State | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Healthy | 100% | 99.60% | 99.51% | 99.57% | 1.08s | 5.03s | 41 tps | 131K | $0.200 / $0.520 | |
| Healthy | 100% | 99.81% | 99.71% | 99.84% | 497ms | 1.54s | 82 tps | 128K | $0.710 / $0.710 | |
PrimeIntellectprimeintellect | Healthy | 100% | 97.93% | 100% | 100% | 665ms | 2.38s | 34 tps | 131K | $1.04 / $2.25 |
Groqgroq | Healthy | 99.96% | 99.82% | 99.80% | 99.78% | 354ms | 1.06s | 234 tps | 131K | $0.590 / $0.790 |
| Healthy | 99.91% | 99.05% | 98.61% | 99.35% | 552ms | 3.03s | 62 tps | 131K | $0.220 / $0.500 | |
| Healthy | 99.85% | 94.76% | 93.58% | 97.47% | 1.01s | 4.10s | 34 tps | 12K | $0.135 / $0.400 | |
| Healthy | 99.72% | 99.18% | 98.81% | 97.94% | 463ms | 5.18s | 37 tps | 24K | $0.293 / $2.25 | |
SambaNovasambanova-turbo | Healthy | 99.62% | 99.25% | 99.13% | 98.85% | 746ms | 3.16s | 116 tps | 131K | $0.450 / $0.900 |
| Healthy | 98.98% | 99.29% | 96.57% | 98.04% | 509ms | 11.4s | 16 tps | 131K | $0.100 / $0.320 | |
Togethertogether | Healthy | 98.15% | 99.01% | 98.40% | 97.98% | 739ms | 3.96s | 48 tps | 131K | $1.04 / $1.04 |
| No data | — | — | — | 99.87% | — | — | — | 131K | $0.250 / $0.750 | |
Googlegoogle-vertex | No data | — | — | — | 99.15% | — | — | — | 128K | $0.720 / $0.720 |
Googlegoogle-vertex/us-central1 | No data | — | — | — | 99.96% | 749ms | 749ms | 141 tps | 128K | $0.720 / $0.720 |
| No data | — | — | — | 56.10% | — | — | — | 131K | $0.130 / $0.400 |
- AkashML100%Healthy24h 99.60%
- 7d
- 99.51%
- 30d
- 99.57%
- p50
- 1.08s
- p99
- 5.03s
- Throughput
- 41 tps
- Context
- 131K
- $/Mtok
- $0.200 / $0.520
akashml/fp8fp8 - CoreWeave100%Healthy24h 99.81%
- 7d
- 99.71%
- 30d
- 99.84%
- p50
- 497ms
- p99
- 1.54s
- Throughput
- 82 tps
- Context
- 128K
- $/Mtok
- $0.710 / $0.710
coreweave/fp16fp16 - PrimeIntellect100%Healthy24h 97.93%
- 7d
- 100%
- 30d
- 100%
- p50
- 665ms
- p99
- 2.38s
- Throughput
- 34 tps
- Context
- 131K
- $/Mtok
- $1.04 / $2.25
primeintellect - Groq99.96%Healthy24h 99.82%
- 7d
- 99.80%
- 30d
- 99.78%
- p50
- 354ms
- p99
- 1.06s
- Throughput
- 234 tps
- Context
- 131K
- $/Mtok
- $0.590 / $0.790
groq - Parasail99.91%Healthy24h 99.05%
- 7d
- 98.61%
- 30d
- 99.35%
- p50
- 552ms
- p99
- 3.03s
- Throughput
- 62 tps
- Context
- 131K
- $/Mtok
- $0.220 / $0.500
parasail/fp8fp8 - Novita99.85%Healthy24h 94.76%
- 7d
- 93.58%
- 30d
- 97.47%
- p50
- 1.01s
- p99
- 4.10s
- Throughput
- 34 tps
- Context
- 12K
- $/Mtok
- $0.135 / $0.400
novita/bf16bf16 - Cloudflare99.72%Healthy24h 99.18%
- 7d
- 98.81%
- 30d
- 97.94%
- p50
- 463ms
- p99
- 5.18s
- Throughput
- 37 tps
- Context
- 24K
- $/Mtok
- $0.293 / $2.25
cloudflare/fp8fp8 - SambaNova99.62%Healthy24h 99.25%
- 7d
- 99.13%
- 30d
- 98.85%
- p50
- 746ms
- p99
- 3.16s
- Throughput
- 116 tps
- Context
- 131K
- $/Mtok
- $0.450 / $0.900
sambanova-turbo - DeepInfra98.98%Healthy24h 99.29%
- 7d
- 96.57%
- 30d
- 98.04%
- p50
- 509ms
- p99
- 11.4s
- Throughput
- 16 tps
- Context
- 131K
- $/Mtok
- $0.100 / $0.320
deepinfra/turbofp8 - Together98.15%Healthy24h 99.01%
- 7d
- 98.40%
- 30d
- 97.98%
- p50
- 739ms
- p99
- 3.96s
- Throughput
- 48 tps
- Context
- 131K
- $/Mtok
- $1.04 / $1.04
together - No data24h —
- 7d
- —
- 30d
- 99.87%
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 131K
- $/Mtok
- $0.250 / $0.750
crusoe/bf16bf16 - No data24h —
- 7d
- —
- 30d
- 99.15%
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 128K
- $/Mtok
- $0.720 / $0.720
google-vertex - No data24h —
- 7d
- —
- 30d
- 99.96%
- p50
- 749ms
- p99
- 749ms
- Throughput
- 141 tps
- Context
- 128K
- $/Mtok
- $0.720 / $0.720
google-vertex/us-central1 - No data24h —
- 7d
- —
- 30d
- 56.10%
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 131K
- $/Mtok
- $0.130 / $0.400
nebius/fp8fp8
Measured 4m ago. Latency and throughput are OpenRouter's rolling 30-minute windows.
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.
Daily availability
One row per endpoint · one dot per day · last 90 days.
worst hour ≥ 95%90–95%< 90%no data
Colour by each endpoint's worst hour of each UTC day · 14 endpoints · 47 days with data, 104 endpoint-days below 90%.
Price changes
- Llama 3.3 70B InstructAkashMLakashml/fp8$0.130 → $0.200 / $0.400 → $0.520cache read — → $0.1002026-08-19 09:50 UTC