Home › Models › Qwen2.5 72B

Qwen2.5 72B

by Alibaba Cloud · Text & Chat · Code

Top open-source model from Alibaba.

85/100AIVerso score
#25 of 49in Text & Chat
Free (Open Source)Starting price
YesFree tier
Try Qwen2.5 72B →Compare with alternatives →

At a glance

ProviderAlibaba Cloud
CategoryText & Chat, Code
Starting priceFree (Open Source)
Free tierYes
API accessYes
Open sourceYes
Context window128K tokens
Mobile appNo

Key facts

Capability ratings

Math & Logice.g. AIME, MATH-500
78
Programminge.g. SWE-bench, LiveCodeBench
75
Reasoninge.g. GPQA, ARC
78
Knowledgee.g. MMLU-Pro
80
Long Contexte.g. MRCR, needle tests
80
Instruction Followinge.g. IFEval
80
Agent Capabilitye.g. Terminal-Bench, tool use
64

AIVerso capability ratings (0–100), an editorial summary informed by public benchmark results. Higher is better.

Strengths

Free tier available
Open source / open weights — self-hostable
API access for developers

Limitations

No official mobile app

Best alternatives to Qwen2.5 72B

Claude Fable 5.1
Anthropic · $10/$50 per 1M tokens — Higher score (99 vs 85)
99
GPT-6 Astra
OpenAI · From $20/mo (Plus) — Higher score (98 vs 85)
98
Fugu Ultra v2
Sakana AI · $5/$30 per 1M tokens — Higher score (97 vs 85)
97
Gemini 3.1 Pro
Google · Free – $20/mo · free tier — Higher score (96 vs 85)
96
MiMo-V2.6-Pro
Xiaomi · Free (open weights) · $0.44/$0.87 per 1M · free tier — Higher score (96 vs 85)
96

🔗 Comparisons & guides

❓ Frequently Asked Questions

Is Qwen2.5 72B free?

Yes — Qwen2.5 72B has a free tier. Pricing: Free (Open Source).

What are the best alternatives to Qwen2.5 72B?

The highest-rated AI chatbots & text models from other providers on AIVerso are Claude Fable 5.1 (99/100), GPT-6 Astra (98/100), Fugu Ultra v2 (97/100).

Does Qwen2.5 72B have an API?

Yes, Qwen2.5 72B can be used via an API.

Is Qwen2.5 72B open source?

Yes — Qwen2.5 72B is available as open source / open weights, so it can be self-hosted.

How does AIVerso rate Qwen2.5 72B?

Qwen2.5 72B scores 85/100 on AIVerso and ranks #25 of 49 in Text & Chat. Scores combine public benchmark results, features and pricing; commissions never influence them.

How we rate: Scores and capability ratings are based on public benchmarks, provider documentation, features and pricing — not on commission rates. Prices can change; check the provider's site before buying. Last updated: September 29, 2026.