token.app › Compare › Qwen2.5 72B Instruct vs Qwen3.5-Flash

Qwen2.5 72B Instruct vs Qwen3.5-Flash

Qwen3.5-Flash leads on 2 of 2 shared benchmarks. Qwen3.5-Flash is 3.3× cheaper per token on a blended 3:1 basis. Qwen3.5-Flash accepts the larger context window (1M).

Head to head
Blended price uses a 3:1 input:output mix
Qwen2.5 72B InstructQwen3.5-Flash
ProviderAlibabaAlibaba
Input $/1M$0.360$0.065
Output $/1M$0.400$0.260
Blended $/1M$0.370$0.114
Context window33K1M
Max output16K66K
ReleasedSep 2024Feb 2026
LicenceOpen weightsProprietary
GPQA Diamond 49.1% 82.3%
OTIS Mock AIME 8.1% 84.4%
Benchmark scores independently run and published by Epoch AI, 'AI Benchmarking Hub', used under CC BY 4.0. Only benchmarks with a score for BOTH models are compared.
Choose Qwen2.5 72B Instruct if…
  • Open weights — self-hostable
Choose Qwen3.5-Flash if…
  • Costs less per token — $0.114 vs $0.370 blended
  • Higher GPQA Diamond (82.3% vs 49.1%)
  • Higher OTIS Mock AIME (84.4% vs 8.1%)
  • Larger context window (1M)
  • Newer model (released Feb 2026)
FAQ
Which is better, Qwen2.5 72B Instruct or Qwen3.5-Flash?
Across the 2 benchmarks both models have independently-run scores for, Qwen2.5 72B Instruct leads on 0 and Qwen3.5-Flash leads on 2. Benchmarks are one input — price and context window are on this page too.
Is Qwen2.5 72B Instruct cheaper than Qwen3.5-Flash?
Qwen3.5-Flash is cheaper: $0.114 versus $0.370 per million tokens blended at 3:1 input:output. Input alone: $0.360 vs $0.065. Output alone: $0.400 vs $0.260.
What context windows do Qwen2.5 72B Instruct and Qwen3.5-Flash have?
Qwen2.5 72B Instruct: 32,768 tokens. Qwen3.5-Flash: 1,000,000 tokens.
Who makes Qwen2.5 72B Instruct and Qwen3.5-Flash?
Qwen2.5 72B Instruct is made by Alibaba. Qwen3.5-Flash is made by Alibaba.
Related comparisons
Qwen2.5 72B Instruct full details Qwen3.5-Flash full details Qwen2.5 72B Instruct vs Flint (free) Qwen2.5 72B Instruct vs MiniMax M3.1 Flash (free) Qwen2.5 72B Instruct vs DeepSeek V4.1 Flash 1017 Qwen2.5 72B Instruct vs Glimpse Qwen2.5 72B Instruct vs Xing4.0 29B A4B (free) Qwen2.5 72B Instruct vs Beam 1.0 501B A23B