Route order log in sign up
1 pick your primary route

Qwen: Qwen3.8 27B

qwen/qwen3.8-27b · agent · text+image+video->text · 1M ctx · added 2026-08-14
$0.0585·1.8395 /M effective hit 79% 123.09B/d

Qwen3.8 27B is an open-weight dense vision-language model from Qwen. It is suited for coding, professional workflows, research, multimodal interaction, and long-running agent tasks, with flexible thinking that can be...

view
openrouter/qwen/qwen3.8-27b
Qwen: Qwen3.8 27B · agent · text+image+video->text · 1M ctx
123.09B/d 79% $0.0585·1.8395 – 2mo
sticker
life 429 tok/d dp age
¢ᵉ
chutes/fp8
$0.0390·1.2500·0.0039 $0.0076·1.2500ⁱ $0.0120·1.2492ᵉ 75% – – 4.9% ◐ 2067ms 36t/s 2mo ·
near-ai/fp8
$0.0400·1.3500·0.0180 $0.0203·1.3500ⁱ $0.0236·1.3489ᵉ 75% – – 6.1% ✓ 1665ms 75t/s 1d ·
¢ ✓
darkbloom/fp4
$0.0350·1.5000·0.0175 $0.0193·1.5000ⁱ $0.0319·1.4997ᵉ 21% – – 0.3% ◐ 1800ms 25t/s 2mo ·
≈
nebius/fp4
$0.4500·3.0000·0.1000 $0.1367·3.0000ⁱ $0.0701·1.4188ᵉ 88% – – 21% ✓ 649ms 227t/s 1d ·
streamlake
$0.1750·1.0500·0.0350 $0.0497·1.0500ⁱ $0.1217·1.0496ᵉ 39% – – <0.1% ◐ 7888ms 25t/s 1d ·
¢ⁱ ≈
primeintellect/fp8
$0.1220·0.8670·0.0140 $0.0253·0.8670ⁱ $0.1247·1.0614ᵉ 22% – – 0.4% ✓ 562ms 73t/s 1d ·
ionstream/fp8
$0.0400·1.9500·0.0200 $0.0221·1.9500ⁱ $0.0271·1.9491ᵉ 72% – – 0.7% ✓ 811ms 65t/s 2mo ·
♻ ★
deepinfra/bf16
$0.1500·1.8750·0.0375 $0.0493·1.8750ⁱ $0.0475·1.8745ᵉ 90% – – 34% ✓ 1045ms 20t/s 1mo ·
mancer/fp8
$0.1500·1.2500·0.0750 $0.0829·1.2500ⁱ $0.1237·1.2490ᵉ 38% – – 0.3% ✓ 836ms 69t/s 2mo ·
tenstorrent/fp8
$0.1500·1.6000·0.0500 $0.0605·1.6000ⁱ $0.0894·1.5996ᵉ 69% – – 1.1% ✓ 608ms 109t/s 1d ·
modal
$0.0400·2.3000·0.0200 $0.0221·2.3000ⁱ $0.0281·2.2993ᵉ 62% – – 10% ✓ 849ms 83t/s 1d ·
dekallm
$0.0400·2.5000·0.0200 $0.0221·2.5000ⁱ $0.0279·2.4996ᵉ 64% – – 1.6% ✓ 755ms 121t/s 1mo ·
wafer/fp8
$0.0600·2.2000·0.0590 $0.0591·2.2000ⁱ $0.0592·2.2466ᵉ 75% – – 5.3% ✓ 676ms 132t/s 1d ·
phala
$0.2000·2.5000·0.0500 $0.0657·2.5000ⁱ $0.0730·2.4998ᵉ 81% – – 1.5% ✓ 685ms 104t/s ✕14h ·
parasail/fp8
$0.2400·2.2000·0.0500 $0.0699·2.2000ⁱ $0.1081·2.1990ᵉ 69% – – 2.1% ✓ 642ms 56t/s 2mo ·
akashml/fp8
$0.2250·1.9800·0.0500 $0.0684·1.9800ⁱ $0.1591·1.9791ᵉ 39% – – 0.4% ✓ 773ms 46t/s 2mo ·
reka/fp8
$0.1500·2.5000·0.0500 $0.0605·2.5000ⁱ $0.1211·2.4993ᵉ 27% – – 1.1% ✓ 1450ms 57t/s 2mo ·
modelrun/fp4
$0.2000·2.5000·0.0900 $0.1015·2.5000ⁱ $0.1436·2.4998ᵉ 57% – – 2.0% ✓ 311ms 151t/s ✕14h ·
novita
$0.4200·3.0000·0.0850 $0.1202·3.0000ⁱ $0.1212·3.0000ᵉ 87% – – 0.4% ✓ 2817ms 27t/s 2mo ·
alibaba
$0.4250·2.5500·0.0850 $0.1207·2.5500ⁱ $0.1968·2.5495ᵉ 82% – – 5.8% ◐ 3211ms 31t/s 2mo ·
io-net/fp8
$0.3000·2.8000·0.1800 · $0.1926·2.8000ⁱ $0.2822·2.7996ᵉ · 15% · – – – · ✓ 836ms · 23t/s · ✕1mo ·
coreweave/fp8
$0.4000·3.0000·0.1500 $0.1762·3.0000ⁱ $0.3760·3.0000ᵉ 19% – – 0.5% ✓ 1401ms 28t/s 2mo ·
gmicloud/bf16
$0.4500·3.2000·0.2500 $0.2710·3.2000ⁱ $0.4052·3.1991ᵉ 22% – – <0.1% ◐ 1077ms 62t/s 1d ·
cloudflare
$0.4500·3.2000·0.0500 $0.0920·3.2000ⁱ $0.4499·3.1991ᵉ 0% – – <0.1% ✓ 1102ms 35t/s 2mo ·
venice/fp8
$0.4500·3.2000 $0.4500·3.2000ⁱ $0.4499·3.1994ᵉ 8% – – <0.1% ✓ 950ms 62t/s 2mo ·
⚡
cerebras/fp16
$0.9900·1.4900·0.9900 $0.9900·1.4900ⁱ $0.9899·1.4894ᵉ 48% – – 0.6% ✓ 312ms 836t/s 1d ·
akashml/bf16
$0.4000·2.5500·0.0500 · $0.0867·2.5500ⁱ – · – · – – – · · 2314ms · 27t/s · ✕2mo ·
vercel/alibaba/qwen3.8-27b
Qwen3.8 27B · agent · text+image+pdf+video->text · 262k ctx
– 77% $0.1088·2.2000 – 2mo
requesty/alibaba/qwen3.8-27b
qwen3.8-27b · agent · text+image->text · 262k ctx
– $0.0657·2.5000 – 2mo
hf/Qwen/Qwen3.8-27B
Qwen3.8-27B · agent · text+image->text · 1M ctx
– $0.4200·2.5000 – 2mo
compare all models →