Gemma 4 31B

google/gemma-4-31b-it

Compare with Gemma 4 26B A4B , Gemma 4 26B A4B (free), Gemma 4 31B (free)

CompareGemma 4 31B

Provider	State	Uptime 5m	24h	p50 latency	p99	Throughput	Context	$/Mtok
Cerebrascerebras/fp16fp16	Healthy	100%	100%	377ms	5.56s	115 tps	131K	$0.990 / $1.49
Friendlifriendli	Healthy	100%	99.83%	4.03s	11.6s	19 tps	262K	$0.140 / $0.400
DeepInfradeepinfra/turbofp4	Healthy	99.91%	99.11%	511ms	4.35s	35 tps	262K	$0.090 / $0.340
Venicevenice/bf16bf16	Healthy	99.37%	99.61%	910ms	16.3s	25 tps	256K	$0.120 / $0.360
Parasailparasail/fp8fp8	Healthy	99.08%	99.09%	975ms	16.1s	16 tps	262K	$0.150 / $0.400
Togethertogether	Healthy	98.80%	92.29%	848ms	9.76s	17 tps	262K	$0.390 / $0.970
OpenInferenceopen-inference/bf16bf16	Healthy	98.41%	95.61%	1.64s	25.1s	12 tps	262K	$0.080 / $0.350
CoreWeavecoreweave/bf16bf16	Healthy	98.23%	98.20%	964ms	21.8s	24 tps	262K	$0.100 / $0.340
Crusoecrusoe	Healthy	97.36%	98.99%	950ms	17.7s	19 tps	262K	$0.140 / $0.400
DeepInfradeepinfra/ultrafp8	Healthy	97.30%	74.41%	4.76s	12.3s	18 tps	131K	$0.270 / $0.760
Phalaphala	Degraded	94.97%	95.31%	3.83s	121s	16 tps	262K	$0.150 / $0.460
SiliconFlowsiliconflow/fp8fp8	Degraded	93.81%	95.90%	2.50s	16.2s	14 tps	262K	$0.130 / $0.400
ModelRunmodelrun/fp4fp4	Down	62.43%	98.93%	240ms	20.2s	65 tps	262K	$0.750 / $1.00
Novitanovita/bf16bf16	Down	46.06%	62.63%	5.00s	27.4s	1.0 tps	262K	$0.140 / $0.400
Chuteschutes/fp4fp4	Down	4.65%	90.44%	21.9s	160s	2.5 tps	131K	$0.120 / $0.370
DeepInfradeepinfra/fp8fp8	Down	—	94.46%	5.67s	25.7s	1.0 tps	262K	$0.130 / $0.380
SambaNovasambanova	Degraded	—	97.43%	1.64s	10.9s	50 tps	262K	$0.380 / $1.15

Cerebras100%
Healthy24h 100%
p50
377ms
p99
5.56s
Throughput
115 tps
Context
131K
$/Mtok
$0.990 / $1.49
cerebras/fp16fp16
Friendli100%
Healthy24h 99.83%
p50
4.03s
p99
11.6s
Throughput
19 tps
Context
262K
$/Mtok
$0.140 / $0.400
friendli
DeepInfra99.91%
Healthy24h 99.11%
p50
511ms
p99
4.35s
Throughput
35 tps
Context
262K
$/Mtok
$0.090 / $0.340
deepinfra/turbofp4
Venice99.37%
Healthy24h 99.61%
p50
910ms
p99
16.3s
Throughput
25 tps
Context
256K
$/Mtok
$0.120 / $0.360
venice/bf16bf16
Parasail99.08%
Healthy24h 99.09%
p50
975ms
p99
16.1s
Throughput
16 tps
Context
262K
$/Mtok
$0.150 / $0.400
parasail/fp8fp8
Together98.80%
Healthy24h 92.29%
p50
848ms
p99
9.76s
Throughput
17 tps
Context
262K
$/Mtok
$0.390 / $0.970
together
OpenInference98.41%
Healthy24h 95.61%
p50
1.64s
p99
25.1s
Throughput
12 tps
Context
262K
$/Mtok
$0.080 / $0.350
open-inference/bf16bf16
CoreWeave98.23%
Healthy24h 98.20%
p50
964ms
p99
21.8s
Throughput
24 tps
Context
262K
$/Mtok
$0.100 / $0.340
coreweave/bf16bf16
Crusoe97.36%
Healthy24h 98.99%
p50
950ms
p99
17.7s
Throughput
19 tps
Context
262K
$/Mtok
$0.140 / $0.400
crusoe
DeepInfra97.30%
Healthy24h 74.41%
p50
4.76s
p99
12.3s
Throughput
18 tps
Context
131K
$/Mtok
$0.270 / $0.760
deepinfra/ultrafp8
Phala94.97%
Degraded24h 95.31%
p50
3.83s
p99
121s
Throughput
16 tps
Context
262K
$/Mtok
$0.150 / $0.460
phala
SiliconFlow93.81%
Degraded24h 95.90%
p50
2.50s
p99
16.2s
Throughput
14 tps
Context
262K
$/Mtok
$0.130 / $0.400
siliconflow/fp8fp8
ModelRun62.43%
Down24h 98.93%
p50
240ms
p99
20.2s
Throughput
65 tps
Context
262K
$/Mtok
$0.750 / $1.00
modelrun/fp4fp4
Novita46.06%
Down24h 62.63%
p50
5.00s
p99
27.4s
Throughput
1.0 tps
Context
262K
$/Mtok
$0.140 / $0.400
novita/bf16bf16
Chutes4.65%
Down24h 90.44%
p50
21.9s
p99
160s
Throughput
2.5 tps
Context
131K
$/Mtok
$0.120 / $0.370
chutes/fp4fp4
DeepInfra—
Down24h 94.46%
p50
5.67s
p99
25.7s
Throughput
1.0 tps
Context
262K
$/Mtok
$0.130 / $0.380
deepinfra/fp8fp8
SambaNova—
Degraded24h 97.43%
p50
1.64s
p99
10.9s
Throughput
50 tps
Context
262K
$/Mtok
$0.380 / $1.15
sambanova

Measured 37s 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 and p99 are OpenRouter's rolling 30-minute windows, averaged over the hour.

Throughput

Median tokens per second.