Llama 3.3 70B 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
AkashMLakashml/fp8fp8	Healthy	100%◆ (best)	96.97%	97.25%	97.25%	776ms	5.62s	21 tps	131K	$0.200 / $0.520
CoreWeavecoreweave/fp16fp16	Healthy	100%◆ (best)	98.60%	97.79%	97.79%	155ms◆ (best)	1.54s	38 tps	128K	$0.710 / $0.710
Crusoecrusoe/bf16bf16	Healthy	100%◆ (best)	99.74%	99.07%	99.07%	366ms	2.58s	53 tps	131K	$0.250 / $0.750
Groqgroq	Healthy	100%◆ (best)	99.91%	99.56%	99.56%	214ms	724ms◆ (best)	16 tps	131K	$0.590 / $0.790
SambaNovasambanova-turbo	Healthy	99.64%	96.74%	95.40%	95.40%	584ms	5.33s	105 tps◆ (best)	131K	$0.450 / $0.900
Parasailparasail/fp8fp8	Healthy	98.64%	98.29%	97.89%	97.89%	675ms	5.94s	37 tps	131K	$0.220 / $0.500
Novitanovita/bf16bf16	Healthy	98.54%	94.81%	94.26%	94.26%	774ms	4.18s	29 tps	12K	$0.135 / $0.400
Cloudflarecloudflare/fp8fp8	Healthy	97.37%	95.99%	94.52%	94.52%	741ms	6.63s	29 tps	24K	$0.293 / $2.25
DeepInfradeepinfra/turbofp8	Degraded	93.33%	96.87%	96.51%	96.51%	914ms	8.25s	8.0 tps	131K	$0.100 / $0.320◆ (best)
Googlegoogle-vertex	No data	—	98.83%	98.27%	98.27%	377ms	2.50s	35 tps	128K	$0.720 / $0.720
Googlegoogle-vertex/us-central1	No data	—	99.99%	100%◆ (best)	100%◆ (best)	362ms	2.37s	46 tps	128K	$0.720 / $0.720
Nebiusnebius/fp8fp8	No data	—	40.94%	42.81%	42.81%	4.62s	28.7s	1.0 tps	131K	$0.130 / $0.400
Togethertogether	No data	—	98.44%	97.80%	97.80%	1.38s	6.62s	14 tps	131K	$1.04 / $1.04

AkashML100%◆ (best)
Healthy24h 96.97%
7d
97.25%
30d
97.25%
p50
776ms
p99
5.62s
Throughput
21 tps
Context
131K
$/Mtok
$0.200 / $0.520
akashml/fp8fp8
CoreWeave100%◆ (best)
Healthy24h 98.60%
7d
97.79%
30d
97.79%
p50
155ms◆ (best)
p99
1.54s
Throughput
38 tps
Context
128K
$/Mtok
$0.710 / $0.710
coreweave/fp16fp16
Crusoe100%◆ (best)
Healthy24h 99.74%
7d
99.07%
30d
99.07%
p50
366ms
p99
2.58s
Throughput
53 tps
Context
131K
$/Mtok
$0.250 / $0.750
crusoe/bf16bf16
Groq100%◆ (best)
Healthy24h 99.91%
7d
99.56%
30d
99.56%
p50
214ms
p99
724ms◆ (best)
Throughput
16 tps
Context
131K
$/Mtok
$0.590 / $0.790
groq
SambaNova99.64%
Healthy24h 96.74%
7d
95.40%
30d
95.40%
p50
584ms
p99
5.33s
Throughput
105 tps◆ (best)
Context
131K
$/Mtok
$0.450 / $0.900
sambanova-turbo
Parasail98.64%
Healthy24h 98.29%
7d
97.89%
30d
97.89%
p50
675ms
p99
5.94s
Throughput
37 tps
Context
131K
$/Mtok
$0.220 / $0.500
parasail/fp8fp8
Novita98.54%
Healthy24h 94.81%
7d
94.26%
30d
94.26%
p50
774ms
p99
4.18s
Throughput
29 tps
Context
12K
$/Mtok
$0.135 / $0.400
novita/bf16bf16
Cloudflare97.37%
Healthy24h 95.99%
7d
94.52%
30d
94.52%
p50
741ms
p99
6.63s
Throughput
29 tps
Context
24K
$/Mtok
$0.293 / $2.25
cloudflare/fp8fp8
DeepInfra93.33%
Degraded24h 96.87%
7d
96.51%
30d
96.51%
p50
914ms
p99
8.25s
Throughput
8.0 tps
Context
131K
$/Mtok
$0.100 / $0.320◆ (best)
deepinfra/turbofp8
Google—
No data24h 98.83%
7d
98.27%
30d
98.27%
p50
377ms
p99
2.50s
Throughput
35 tps
Context
128K
$/Mtok
$0.720 / $0.720
google-vertex
Google—
No data24h 99.99%
7d
100%◆ (best)
30d
100%◆ (best)
p50
362ms
p99
2.37s
Throughput
46 tps
Context
128K
$/Mtok
$0.720 / $0.720
google-vertex/us-central1
Nebius—
No data24h 40.94%
7d
42.81%
30d
42.81%
p50
4.62s
p99
28.7s
Throughput
1.0 tps
Context
131K
$/Mtok
$0.130 / $0.400
nebius/fp8fp8
Together—
No data24h 98.44%
7d
97.80%
30d
97.80%
p50
1.38s
p99
6.62s
Throughput
14 tps
Context
131K
$/Mtok
$1.04 / $1.04
together

Measured 2m 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.