Skip to content

Qwen3.5-35B-A3B

text + image + video → textHugging Face OpenRouter Cost calculator →

Best price · blended / 1M

$0.236

at Darkbloom

Input / output
$0.065 / $0.75
Providers
7
Spread
2.4×
7 days
0%
30 days
0%
Context
256K
Prices per 1M tokens, blended: Darkbloom $0.236, DeepInfra (fp8/flex) $0.284, DeepInfra $0.355, Parasail $0.363, Venice $0.363, Alibaba $0.447, AtlasCloud $0.619, SiliconFlow $0.63.

Providers

  • DarkbloomCheapest
    $0.236
    $0.065 in · $0.75 out
    −9.2% vs ref
    256K ctx99.7% uptimefp4
  • fp8/flex
    $0.284
    $0.112 in · $0.80 out
    +9.2% vs ref
    256K ctx0.0% uptimefp8
  • $0.355
    $0.14 in · $1.00 out
    +36% vs ref
    256K ctx98.5% uptimefp8
  • $0.363
    $0.15 in · $1.00 out
    +39% vs ref
    256K ctx99.9% uptimefp8
  • $0.363
    $0.15 in · $1.00 out
    +39% vs ref
    250K ctx99.9% uptime
  • $0.447
    $0.163 in · $1.30 out
    +72% vs ref
    256K ctx100% uptime
  • $0.619
    $0.225 in · $1.80 out
    +138% vs ref
    256K ctx93.3% uptimefp8
  • $0.63
    $0.24 in · $1.80 out
    +142% vs ref
    256K ctx73.5% uptimefp8
Blended = (3 × input + output) / 4. Ref $0.26 = median of cheapest 20%.

Price history

USD per 1M tokens · each step is a price change

Prices in USD / 1M tokens.

Other pricing

Offered byCache read
Darkbloom$0.0325
DeepInfra (fp8/flex)$0.04
DeepInfra, Parasail +1$0.05
AtlasCloud$0.225
SiliconFlow$0.15

Changes

  • Now served,
    Listed
  • Cache read price,
    $0.04→$0.0325−18.8%
  • Input price,
    $0.08→$0.065−18.8%
  • Cache read price,
    $0.156→$0.05−68.0%
  • Output price,
    $1.25→$1.00−20.0%
  • Input price,
    $0.313→$0.15−52.0%

About

The cheapest provider for Qwen3.5-35B-A3B is Darkbloom at $0.236 per 1M tokens (blended), charging $0.065 per 1M input and $0.75 per 1M output tokens. 7 providers serve it; the most expensive charges 2.4× the typical price. Context window: 256K tokens. Prices tracked since Oct 3, 2026.

The Qwen3.5 Series 35B-A3B is a native vision-language model designed with a hybrid architecture that integrates linear attention mechanisms and a sparse mixture-of-experts model, achieving higher inference efficiency. Its overall...
Released
Feb 25, 2026
First tracked
Oct 3, 2026
Canonical slug
qwen/qwen3.5-35b-a3b-20260224
Weights
Qwen/Qwen3.5-35B-A3B