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

	State
Cerebrascerebras/fp16fp16	Healthy	100%	100%	99.99%	99.99%	259ms	1.92s	146 tps	131K	$0.990 / $1.49
DeepInfradeepinfra/ultrafp8	Healthy	100%	74.48%	87.73%	87.73%	2.29s	8.19s	19 tps	131K	$0.270 / $0.760
ModelRunmodelrun/fp4fp4	Healthy	100%	99.93%	99.64%	99.64%	196ms	3.34s	91 tps	262K	$0.750 / $1.00
SiliconFlowsiliconflow/fp8fp8	Healthy	100%	74.92%	93.86%	93.86%	1.44s	16.5s	25 tps	262K	$0.130 / $0.400
Togethertogether	Healthy	100%	89.68%	95.30%	95.30%	363ms	11.2s	33 tps	262K	$0.390 / $0.970
DeepInfradeepinfra/turbofp4	Healthy	99.95%	99.84%	99.51%	99.51%	559ms	3.56s	44 tps	262K	$0.090 / $0.340
Friendlifriendli	Healthy	99.91%	99.20%	99.55%	99.55%	711ms	6.23s	86 tps	262K	$0.140 / $0.400
Crusoecrusoe	Healthy	99.78%	98.77%	98.60%	98.60%	545ms	14.2s	35 tps	262K	$0.140 / $0.400
Venicevenice/bf16bf16	Healthy	99.69%	99.74%	99.73%	99.73%	1.31s	12.9s	37 tps	256K	$0.120 / $0.360
DeepInfradeepinfra/fp8fp8	Healthy	99.66%	98.33%	90.43%	90.43%	689ms	8.07s	46 tps	262K	$0.130 / $0.380
Parasailparasail/fp8fp8	Healthy	98.53%	99.49%	99.27%	99.27%	907ms	10.4s	39 tps	262K	$0.150 / $0.400
Novitanovita/bf16bf16	Healthy	98.20%	99.08%	78.32%	78.32%	904ms	3.29s	9.0 tps	262K	$0.140 / $0.400
OpenInferenceopen-inference/bf16bf16	Healthy	97.95%	97.00%	97.44%	97.44%	878ms	13.4s	26 tps	262K	$0.080 / $0.350
CoreWeavecoreweave/bf16bf16	Healthy	95.63%	98.17%	98.52%	98.52%	579ms	21.5s	30 tps	262K	$0.100 / $0.340
Chuteschutes/fp4fp4	Down	56.41%	86.39%	63.05%	63.05%	10.1s	112s	5.0 tps	131K	$0.120 / $0.370
Phalaphala	Down	0.00%	81.12%	88.05%	88.05%	—	—	—	262K	$0.150 / $0.460
SambaNovasambanova	No data	—	98.66%	96.98%	96.98%	1.61s	9.55s	22 tps	131K	$0.380 / $1.15

Cerebras100%
Healthy24h 100%
7d
99.99%
30d
99.99%
p50
259ms
p99
1.92s
Throughput
146 tps
Context
131K
$/Mtok
$0.990 / $1.49
cerebras/fp16fp16
DeepInfra100%
Healthy24h 74.48%
7d
87.73%
30d
87.73%
p50
2.29s
p99
8.19s
Throughput
19 tps
Context
131K
$/Mtok
$0.270 / $0.760
deepinfra/ultrafp8
ModelRun100%
Healthy24h 99.93%
7d
99.64%
30d
99.64%
p50
196ms
p99
3.34s
Throughput
91 tps
Context
262K
$/Mtok
$0.750 / $1.00
modelrun/fp4fp4
SiliconFlow100%
Healthy24h 74.92%
7d
93.86%
30d
93.86%
p50
1.44s
p99
16.5s
Throughput
25 tps
Context
262K
$/Mtok
$0.130 / $0.400
siliconflow/fp8fp8
Together100%
Healthy24h 89.68%
7d
95.30%
30d
95.30%
p50
363ms
p99
11.2s
Throughput
33 tps
Context
262K
$/Mtok
$0.390 / $0.970
together
DeepInfra99.95%
Healthy24h 99.84%
7d
99.51%
30d
99.51%
p50
559ms
p99
3.56s
Throughput
44 tps
Context
262K
$/Mtok
$0.090 / $0.340
deepinfra/turbofp4
Friendli99.91%
Healthy24h 99.20%
7d
99.55%
30d
99.55%
p50
711ms
p99
6.23s
Throughput
86 tps
Context
262K
$/Mtok
$0.140 / $0.400
friendli
Crusoe99.78%
Healthy24h 98.77%
7d
98.60%
30d
98.60%
p50
545ms
p99
14.2s
Throughput
35 tps
Context
262K
$/Mtok
$0.140 / $0.400
crusoe
Venice99.69%
Healthy24h 99.74%
7d
99.73%
30d
99.73%
p50
1.31s
p99
12.9s
Throughput
37 tps
Context
256K
$/Mtok
$0.120 / $0.360
venice/bf16bf16
DeepInfra99.66%
Healthy24h 98.33%
7d
90.43%
30d
90.43%
p50
689ms
p99
8.07s
Throughput
46 tps
Context
262K
$/Mtok
$0.130 / $0.380
deepinfra/fp8fp8
Parasail98.53%
Healthy24h 99.49%
7d
99.27%
30d
99.27%
p50
907ms
p99
10.4s
Throughput
39 tps
Context
262K
$/Mtok
$0.150 / $0.400
parasail/fp8fp8
Novita98.20%
Healthy24h 99.08%
7d
78.32%
30d
78.32%
p50
904ms
p99
3.29s
Throughput
9.0 tps
Context
262K
$/Mtok
$0.140 / $0.400
novita/bf16bf16
OpenInference97.95%
Healthy24h 97.00%
7d
97.44%
30d
97.44%
p50
878ms
p99
13.4s
Throughput
26 tps
Context
262K
$/Mtok
$0.080 / $0.350
open-inference/bf16bf16
CoreWeave95.63%
Healthy24h 98.17%
7d
98.52%
30d
98.52%
p50
579ms
p99
21.5s
Throughput
30 tps
Context
262K
$/Mtok
$0.100 / $0.340
coreweave/bf16bf16
Chutes56.41%
Down24h 86.39%
7d
63.05%
30d
63.05%
p50
10.1s
p99
112s
Throughput
5.0 tps
Context
131K
$/Mtok
$0.120 / $0.370
chutes/fp4fp4
Phala0.00%
Down24h 81.12%
7d
88.05%
30d
88.05%
p50
—
p99
—
Throughput
—
Context
262K
$/Mtok
$0.150 / $0.460
phala
SambaNova—
No data24h 98.66%
7d
96.98%
30d
96.98%
p50
1.61s
p99
9.55s
Throughput
22 tps
Context
131K
$/Mtok
$0.380 / $1.15
sambanova

Measured 2m ago. Latency and throughput are OpenRouter's rolling 30-minute windows.

Price changes

Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 10:05 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 10:00 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 09:55 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 09:50 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 09:45 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 09:40 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 09:35 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 09:30 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 09:25 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 09:20 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 09:15 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 09:10 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 09:05 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 09:00 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 08:55 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 08:50 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 08:45 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 08:40 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 08:35 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 08:30 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 08:25 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 08:20 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 08:15 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 08:10 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 08:05 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 08:00 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 07:55 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 07:50 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 07:45 UTC
Gemma 4 31B Togethertogether$0.390 / $0.970 → $0.280 / $0.8602026-08-21 07:40 UTC

History

1 endpoint reporting no data is 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.