Qwen 2.5 72B
Available alibaba/qwen-2.5-72b · by Alibaba · decoder-only-transformer
Pricing — 1 offering(s)
Together AI
Input tokens
- $0.36 / 1M tokens (input) Current 2024-09-19 → present
Output tokens
- $0.40 / 1M tokens (output) Current 2024-09-19 → present
Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.
Capability profile
reasoning moderate
coding strong
tool use moderate
instruction following strong
context window moderate
multilingual strong
speed moderate
Operator guidance
The best open-weights value in the 70B tier for multilingual or code tasks. Route here for workloads where Llama 3.3 70B is the default but Chinese language quality, coding, or cost is a primary concern. Step up to Qwen 3 235B for harder reasoning at modestly higher cost.
Use cases
- Multilingual workloads needing open-weights and low per-token cost
- Code generation and technical reasoning at budget pricing
- Self-hosted inference where Chinese-language capability is required
Limitations
- Superseded by Qwen 3 family in capability; Qwen 2.5 72B remains relevant for cost
- Chinese lab; data-residency or compliance constraints may apply
- Dense architecture is compute-heavy for self-hosting
- Capability ratings are qualitative, not a single cited benchmark run