Skip to content
Model Pareto

Qwen3 Next 80B A3B Instruct

Alibaba QwenOpen weightsartificialanalysis.ai ↗
Input
$0.15/M tok
Output
$1.10/M tok
Blended
$0.388/M tok · 3:1
Speed
158tok/s
Context
262Ktokens

Long-tail import from Artificial Analysis (2026-09-24); AA variant: Qwen3 Next 80B A3B Instruct. Price = AA median host price.

Rankings

Self-hostingEstimated

How we estimate →

What it would cost to run these open weights yourself on rented GPUs, compared with the API price above.

Hardware
1× B200 180GB
FP8 weights
Throughput
~9,492 tok/s
many requests batched
Price
$0.063–0.56 /M tok
busy → light use
On your own machine
Mac Studio M5 Max 128GB (int4, ~290 tok/s single-stream)
single consumer GPU or Mac
Assumptions (5)
  • FP8 weights (80B params, 3B active per token) + 25% KV-cache headroom ≈ 100 GB
  • 1× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
  • ~9,492 output tok/s aggregate at batch 256 (bandwidth-bound); MoE compute scales with active params
  • Blended 3:1 input:output; prefill ~354,375 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.

Writing & chat

Hard reasoning

Agents

OSWorld 2.0Estimated
~#3 of 5 (estimated)~53.8%
  1. #1GLM-5.3 Flash59.1%
  2. #2Kimi K358.3%
  3. ~#3Qwen3 Next 80B A3B InstructEstimated53.8%
  4. #3GPT-5.6 Terra50.2%
No published score — estimated from related benchmarks
~#60 of 92 (estimated)~703 Elo
  1. #1Claude Opus 5.51866
  2. #2Muse Wonder Full1857
  3. #59Claude Haiku 4.5719
  4. ~#60Qwen3 Next 80B A3B InstructEstimated703
  5. #60MiniCPM5-2B694
No published score — estimated from related benchmarks