Qwen2.5-Math 7B
A 7.6 billion-parameterHow many numbers the model learned during training — the usual rough measure of its size. dense model from Qwen Team (Alibaba Cloud). Math-specialised base and instruct models.
- Size
- 7.62Bdense
- Memory to run
- ~6 GBsmallest, 8K ctx
- Context
- 4Ktokens
- License
- PermissiveApache License 2.0; MIT License
- Updated
- 19 Sept 2024
Which version to use
- Variant DeepSeek-R1-Distill-Qwen-7B · DeepSeekA smaller model trained to imitate a larger one.
Where it runs
Best-fitting version on each reference system, at an 8K contextHow much text the model can consider at once, counted in tokens — roughly ¾ of a word each..
Runs well 12
- A100 80GB server~84 tok/s est.
- Dual RTX 3090~37 tok/s est.
- GB10 mini workstation~11 tok/s est.
- MacBook Pro M3 Max 64 GB~50 tok/s est.
- MacBook Pro M4 Max 128 GB~68 tok/s est.
- Mac mini M4 Pro 48 GB~34 tok/s est.
- Mac Studio M2 Ultra 192 GB~100 tok/s est.
- RTX 4060 Ti 16GB budget build~36 tok/s est.
- RTX 4090 workstation~42 tok/s est.
- RTX 5090 workstation~74 tok/s est.
- RX 7900 XTX desktop~40 tok/s est.
- Ryzen AI Max+ 395 mini PC~11 tok/s est.
Slowly (CPU or offload) 1
- CPU-only Ryzen 9 7950X~10 tok/s est.
Too large 0
None of the reference systems.
Benchmarks
Each benchmarkA fixed set of questions every model is given, so their scores can be compared on the same task. is developer-reported; prompts and settings differ between labs. Bars are relative to the best open result.
Variants & downloads
Each variant is a separate set of weights. Expand one for its downloads: quantizationStoring each of the model’s numbers with fewer bits, so the file is smaller and needs less memory. trades a little quality for a much smaller file, and the memory column adds the working memoryExtra space the model needs while it answers. It grows with the length of the conversation, on top of the file itself. a conversation needs on top.
Variant Qwen2.5-Math 7BPermissive
- What it is
- Pretrained foundation weights — a starting point for fine-tuning, not for chat.
- Publisher
- Qwen Team (Alibaba Cloud)
- License
- Apache License 2.0 · commercial use allowed
- Released
- 19 Sept 2024
- huggingface
- Qwen/Qwen2.5-Math-7B
Contributions are not open yet, so there is nothing here from members.
| Quantization | Format | Bits | Download | Memory @ 8K | Publisher |
|---|---|---|---|---|---|
| Download BF16native | safetensors | 16 | 14.2 GiB | Qwen Team (Alibaba Cloud)Qwen/Qwen2.5-Math-7B |
Variant DeepSeek-R1-Distill-Qwen-7BPermissive
- What it is
- A smaller model trained to imitate a larger one.
- Publisher
- DeepSeek
- License
- MIT License · commercial use allowed
- Released
- 20 Jan 2025
- huggingface
- deepseek-ai/DeepSeek-R1-Distill-Qwen-7B
Contributions are not open yet, so there is nothing here from members.
| Quantization | Format | Bits | Download | Memory @ 8K | Publisher |
|---|---|---|---|---|---|
| Download BF16native | safetensors | 16 | 14.2 GiB | DeepSeekdeepseek-ai/DeepSeek-R1-Distill-Qwen-7B | |
| Download Q4_K_Mk quant | gguf | 4.89 | 4.3 GiB | LM Studio Communitylmstudio-community/DeepSeek-R1-Distill-Qwen-7B-GGUF |
Architecture details
- Parameters
- 7.62B
- Active / token
- All (dense)
- Architecture
- Dense
- Layers
- 28
- Attention heads
- 28
- KV heads
- 4
- Head dim
- 128
- KV cache @ 8K (fp16)
- 0.22 GiB
- Max context
- 4,096 tokens
Measured performance
Throughput on specific systems and runtimes, with the source of each measurement.
Community results
Reviews
Lineage
How this model’s variants relate to each other and to other models.
Sources & history
Sources
- Mutinai illustrative fixtures (illustrative fixture, 1 record, 13 Sept 2026)
External identifiers
- huggingface Qwen/Qwen2.5-Math-7B
- huggingface deepseek-ai/DeepSeek-R1-Distill-Qwen-7B
Timeline
- Qwen2.5 family released