Skip to content
Model Pareto

Qwen-Image-2.1-Turbo

Alibaba QwenOpen weightshuggingface.co ↗
Compare with nearest rivalsCompare with…
Price
$0.016/image
Time
~6.5sper imageest.

8-step distilled 7B, Qwen Research License. Price is Alibaba Cloud Model Studio Singapore list price.

Rankings

Self-hostingEstimated

How we estimate →

What it would cost to run these open weights yourself on rented GPUs. No API sells this model, so this is its price on the chart.

Hardware
1× B200 180GB
BF16 weights
Throughput
~11,593 tok/s
many requests batched
Price
$0.070–0.62 /M tok
busy → light use
On your own machine
RTX 5090 32GB (int4, 300+ tok/s single-stream)
single consumer GPU or Mac
Assumptions (5)
  • BF16 weights (7B params) + 25% KV-cache headroom ≈ 18 GB
  • 1× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
  • ~11,593 output tok/s aggregate at batch 256 (bandwidth-bound)
  • Blended 3:1 input:output; prefill ~75,938 tok/s
  • Low = 75% utilization at the low GPU price; high = 20% at the high price

Benchmark scores

~ italic, dashed = no published score yet, estimated from related benchmarks.

Image generation

AA Text-to-Image ArenaFew votesVerified
#22 of 421028 Elo
  1. #1GPT Image 2.5 Sunburst1196
  2. #2GPT Image 2.5 Flare1190
  3. #21Qwen-Image-2.1Few votes1041
  4. #22Qwen-Image-2.1-TurboFew votes1028
  5. #23FLUX.2 [flex]1025
artificialanalysis.ai · 2026-10-17
LMArena Text-to-ImageEstimated
~#16 of 33 (estimated)~1204 Elo
  1. #1GPT Image 2.5 Sunburst1423
  2. #2GPT Image 2.5 Flare1398
  3. #15Qwen-Image-2.1Few votes1223
  4. ~#16Qwen-Image-2.1-TurboEstimated1204
  5. #16Ideogram 4.0 (Quality)1204
No published score — estimated from related benchmarks
LMArena Image EditEstimated
~#17 of 27 (estimated)~1323 Elo
  1. #1GPT Image 2.5 Sunburst1522
  2. #2GPT Image 2.5 Flare1481
  3. #16Luma UNI 1 Max1334
  4. ~#17Qwen-Image-2.1-TurboEstimated1323
  5. #17Luma UNI 1.11315
No published score — estimated from related benchmarks
GenEvalEstimated
~#3 of 4 (estimated)~85.1%
  1. #1HiDream-O1-Image90.0%
  2. #2Qwen-Image87.0%
  3. ~#3Qwen-Image-2.1-TurboEstimated85.1%
  4. #3Z-Image Turbo82.0%
No published score — estimated from related benchmarks
DPG-BenchEstimated
~#3 of 4 (estimated)~86.6%
  1. #1HiDream-O1-Image89.8%
  2. #2Qwen-Image88.3%
  3. ~#3Qwen-Image-2.1-TurboEstimated86.6%
  4. #3Z-Image Turbo84.9%
No published score — estimated from related benchmarks