Route order log in sign up
1 pick your primary route

Google: Gemma 4 31B

google/gemma-4-31b-it · agent · text+image+video->text · 262k ctx · added 2026-04-02
$0.1260·0.3995 /M effective hit 31% 65.21B/d

Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output. Features a 256K token context window, configurable thinking/reasoning mode, native function...

view
openrouter/google/gemma-4-31b-it
Google: Gemma 4 31B · agent · text+image+video->text · 262k ctx
65.21B/d 31% $0.1260·0.3995 – 7mo
sticker
life 429 tok/d dp age
¢ᵉ
chutes/fp4
$0.1200·0.3700·0.0120 $0.0501·0.3700ⁱ $0.0693·0.3686ᵉ 62% – – 1.8% ◐ 3265ms 21t/s 3mo ·
¢ ¢ⁱ
open-inference/bf16
$0.0800·0.3500·0.0100 · $0.0347·0.3500ⁱ $0.0798·0.3485ᵉ · 0% · – – – · ✓ 1309ms · 15t/s · ✕2mo ·
✓
coreweave/fp4
$0.1000·0.3400·0.1000 $0.1000·0.3400ⁱ $0.0999·0.3386ᵉ 33% – – 22% ✓ 430ms 63t/s 2mo ·
kt
$0.1000·0.3400·0.1000 $0.1000·0.3400ⁱ $0.0999·0.3385ᵉ 18% · – – 2.5% 1648.7M ◐ 487ms 108t/s 11h ·
venice/fp4
$0.1200·0.3600·0.0900 $0.1006·0.3600ⁱ $0.1091·0.3581ᵉ 37% – – 1.9% ✓ 583ms 48t/s 1mo ·
dekallm
$0.1400·0.4000·0.0500 $0.0817·0.4000ⁱ $0.1138·0.3979ᵉ 37% – – 0.2% ✓ 1692ms 20t/s 1d ·
parasail/fp8
$0.1500·0.4000·0.0600 $0.0917·0.4000ⁱ $0.1161·0.3990ᵉ 30% – – 3.5% ✓ 827ms 29t/s 3mo ·
★
deepinfra/turbo
$0.1350·0.5100·0.0750 $0.0961·0.5100ⁱ $0.1082·0.5083ᵉ 47% – – 23% ✓ 542ms 62t/s 3mo ·
nextbit/fp4
$0.1500·0.4000·0.0600 $0.0917·0.4000ⁱ $0.1226·0.3990ᵉ 31% – – 13% ✓ 673ms 44t/s 1d ·
novita/bf16
$0.1400·0.4000 $0.1400·0.4000ⁱ $0.1398·0.3971ᵉ 0% – – 0.9% ✓ 1170ms 21t/s 3mo ·
crusoe/bf16
$0.1400·0.4000·0.1400 $0.1400·0.4000ⁱ $0.1399·0.3990ᵉ 39% – – 2.9% ✓ 686ms 31t/s 1mo ·
friendli
$0.1400·0.4000 $0.1400·0.4000ⁱ $0.1399·0.3992ᵉ 13% – – 12% ◐ 508ms 144t/s 3mo ·
phala
$0.1500·0.4600·0.0750 · $0.1014·0.4600ⁱ $0.1388·0.4586ᵉ · 15% · – – – · ✓ 1031ms · 63t/s · ✕2mo ·
primeintellect/fp4
$0.2000·0.3350·0.1000 $0.1352·0.3350ⁱ $0.1557·0.3344ᵉ 36% – – 0.6% ✓ 579ms 60t/s 1d ·
deepinfra/fp8/flex
$0.1600·0.3200 $0.1600·0.3200ⁱ $0.1599·0.3184ᵉ 0% – – <0.1% ✓ 4082ms 24t/s 1d ·
deepinfra/fp8
$0.2000·0.4000 $0.2000·0.4000ⁱ $0.1999·0.3993ᵉ 0% – – 8.0% ✓ 827ms 26t/s 3mo ·
modal
$0.3400·0.1000·0.1000 $0.1846·0.1000ⁱ $0.2560·0.0996ᵉ 57% – – 1.6% ✓ 991ms 138t/s 1d ·
deepinfra/ultra
$0.2700·0.7600 · $0.2700·0.7600ⁱ $0.2699·0.7574ᵉ · 0% · – – – · ✓ 3885ms · 40t/s · ✕1mo ·
deepinfra/fp8/priority
$0.3000·0.6000 $0.3000·0.6000ⁱ $0.2998·0.5985ᵉ 0% – – 0.7% ✓ 1241ms 24t/s 1d ·
♻
io-net
$0.3610·1.0925·0.1805 $0.2441·1.0925ⁱ $0.3100·1.0873ᵉ 65% – – <0.1% ✓ 878ms 19t/s ✕6h ·
together
$0.3900·0.9700 · $0.3900·0.9700ⁱ $0.3899·0.9690ᵉ · 7% · – – – · ✓ – · – · ✕1mo ·
sambanova
$0.3800·1.1500 $0.3800·1.1500ⁱ $0.3799·1.1495ᵉ 0% – – 0.3% ✓ 1901ms 102t/s 3mo ·
⚡
modelrun/fp4
$0.7500·1.0000·0.2000 $0.3938·1.0000ⁱ $0.5458·1.0000ᵉ 38% – – 4.6% ✓ 158ms 199t/s ✕12h ·
siliconflow/fp8
$0.7500·1.0000·0.2500 $0.4262·1.0000ⁱ $0.6038·1.0000ᵉ 23% – – 0.5% ✓ 1845ms 35t/s 3mo ·
coreweave/bf16
$0.1000·0.3400·0.1000 · $0.1000·0.3400ⁱ – · – · – – – · · 482ms · 32t/s · ✕2mo ·
venice/bf16
$0.1200·0.3600·0.0900 · $0.1006·0.3600ⁱ – · – · – – – · · 634ms · 37t/s · ✕1mo ·
crusoe
$0.1400·0.4000·0.1400 · $0.1400·0.4000ⁱ – · – · – – – · · 1012ms · 11t/s · ✕1mo ·
morph/fp4
$0.1400·0.4000·0.0800 · $0.1011·0.4000ⁱ – · – · – – – · · 7372ms · 11t/s · ✕2mo ·
cerebras/fp16
$0.9900·1.4900·0.9900 · $0.9900·1.4900ⁱ – · – · – – – · · – · – · ✕2mo ·
vercel/google/gemma-4-31b-it
Gemma 4 31B IT · agent · text+image+pdf->text · 262k ctx
– 30% $0.1400·0.4000 – 7mo
requesty/google/gemma-4-31b-it
gemma-4-31b-it · agent · text+image->text · 262k ctx
– $0.0650·0.1900 – 7mo
hf/google/gemma-4-31B-it
gemma-4-31B-it · agent · text+image->text · 262k ctx
– $0.1700·0.4000 – 7mo
compare all models →