Qwen2.5 7B Instruct
alibaba/qwen-2.5-7b-instruct- Alibaba
- Family · Qwen2
- Open weights
- Text → Text
- Tools · JSON
Not on OSMEXO yet
Spec & list price
Context32K
Max output29K
Input / 1M$0.10
Output / 1M$0.202× input
Usage
Tokens / day44.3M▼ 40% 7d
Usage rank#345
Requests / day42.5K
Tool calls / req0.05
Per request · 30 days
Avg input8180.1× median
Avg output1110.2× median
Cache hit56%0.8× median
Tool-call failures0.9%0.5× median
Consumption
Peak 1.7B 24 OctLow 615.8K 30 OctRank today #345
News & trends
News · latest from Alibaba
- Memdoor: A team of 5 Qwen3.8-27B models beats Fable 5 on LiveCodeBench Hardgithub.com · 31 Oct
- Qwen3.8-Flash Now Available in the EU on Alibaba Cloudtwitter.com · 31 Oct
- Alibaba's Qwen3.8-Flash-Next model tried to upload a file to Beijing in a testnotebookcheck.net · 27 Oct
- Neuron Explorer for Qwen3.5-4B: See every neuron at work, make targeted editsgithub.com · 26 Oct
Benchmarks
Benchmark data: Artificial AnalysisOutput speed#23/334145 tok/s2.3× faster than median
First token#50.24 s4.5× quicker than median
Family progress
Intelligence Index
- Family releases7.7
Technical details
- Tokenizer
- Qwen
- Knowledge cutoff
- 2024-06-30
- License
- Apache 2.0
- Parameters
- 7.6B
- Moderation
- none
- Downloads · likes
- 5.9M · 2.5K
Request parameters 14
- tools
- tool_choice
- response_format
- structured_outputs
- temperature
- top_p
- top_k
- min_p
- max_tokens
- stop
+4 more
- seed
- frequency_penalty
- presence_penalty
- repetition_penalty
When to use
Best for
- Low-cost bulk classification and extraction
- Self-hosted lightweight assistants
- Structured JSON output at small scale
- Fine-tuning experiments on an open model
Think twice for
- Hard reasoning or math
- Long document Q&A
- Image understanding
Research papers
- Beyond Successful Play: Diagnosing Opponent Modeling in Large Language Models12 Oct
- Principled Under Pressure: Post-Training Decides Whether LLMs Act on Their Own Moral Judgment6 Oct
- Illusory Truth or Mere Exposure? Model-Dependent Repetition Effects in LLM-Based Social Media Simulations28 Sep
- Aging of Prompt Engineering Techniques Across LLM Versions25 Aug
- Harness Engineering for Predictable Agentic Systems: An Empirical Study of Deterministic Execution Constraints25 Aug
- From Noise to Signal: Improving Security Log Anomaly Detection Using LLMs with Endpoint-Specific Logs20 Aug
- FirstPass: Grounding AI Scientific Judgment in Multi-Round Editorial Outcomes18 Jun
Same family
Qwen2- 6.3
- 6.9
- 7.7
Want Qwen2.5 7B Instruct on OSMEXO?
Sign in to request it. Requests are reviewed; a request does not guarantee availability.