Phi-4-Reasoning is a 14-billion-parameter model trained on top of the base Phi-4 through supervised fine-tuning, using over 1.4 million carefully curated prompts and high-quality responses generated by the o3-mini1 model. The model specializes in tasks requiring complex, multi-step reasoning and demonstrates the ability to generate detailed reasoning chains while efficiently utilizing computational resources during inference.
The model’s performance metrics are impressive: it outperforms significantly larger open models such as DeepSeek-R1-Distill-Llama-70B and approaches the performance level of the full DeepSeek-R1 model across various benchmarks. On the AIME 2025 math problems, Phi-4-Reasoning shows an improvement of more than 50 percentage points compared to the base Phi-4, and on LiveCodeBench programming tasks, it improves by over 25 percentage points. A notable feature of Phi-4-Reasoning is its ability to generalize and transfer knowledge to tasks not explicitly included in the training data, with improvements also observed on general tasks, including instruction following and toxic content detection.
Phi-4-Reasoning is ideally suited for applications that demand reliable logical reasoning under limited computational resources—such as educational platforms for solving mathematical problems, automated programming systems, scientific research tools, and any applications where a balance between high-quality reasoning and resource efficiency is essential.
| Model Name | Context | Type | GPU | Status | Link |
|---|
There are no public endpoints for this model yet.
An endpoint is a server with pre-installed software that automatically downloads neural network weights and launches an OpenAI-compatible API to handle incoming requests.
A private endpoint is created within the user's project, belongs exclusively to the user, and is billed per server configuration in accordance with the infrastructure pricing.
We recommend creating private endpoints when you need to:
| Name | GPU | TPS | Max Concurrency | |||
|---|---|---|---|---|---|---|
32,768.0 tensor |
2 | $0.57 | 23.38 | 1.562 | Launch | |
32,768.0 tensor |
2 | $0.93 | 54.68 | 3.846 | Launch | |
32,768.0 pipeline |
3 | $1.43 | 1.082 | Launch | ||
32,768.0 tensor |
2 | $1.56 | 62.54 | 4.146 | Launch | |
32,768.0 tensor |
2 | $1.56 | 4.146 | Launch | ||
32,768.0 |
1 | $1.59 | 2.050 | Launch | ||
32,768.0 tensor |
2 | $1.92 | 80.98 | 4.147 | Launch | |
32,768.0 |
1 | $2.37 | 83.01 | 9.024 | Launch | |
32,768.0 |
1 | $3.83 | 84.90 | 8.950 | Launch | |
32,768.0 |
1 | $4.11 | 147.58 | 11.000 | Launch | |
32,768.0 tensor |
2 | $4.61 | 20.554 | Launch | ||
32,768.0 |
1 | $4.74 | 18.014 | Launch | ||
32,768.0 tensor |
2 | $9.40 | 38.380 | Launch | ||
| Name | GPU | TPS | Max Concurrency | |||
|---|---|---|---|---|---|---|
32,768.0 tensor |
2 | $0.93 | 1.773 | Launch | ||
32,768.0 pipeline |
3 | $1.06 | 5.77 | 1.091 | Launch | |
32,768.0 tensor |
2 | $1.56 | 48.66 | 2.085 | Launch | |
32,768.0 tensor |
2 | $1.56 | 2.085 | Launch | ||
32,768.0 tensor |
2 | $1.92 | 58.27 | 2.073 | Launch | |
32,768.0 |
1 | $2.37 | 48.84 | 7.035 | Launch | |
32,768.0 tensor |
2 | $2.93 | 4.378 | Launch | ||
32,768.0 |
1 | $3.83 | 57.95 | 7.025 | Launch | |
32,768.0 |
1 | $4.11 | 9.072 | Launch | ||
32,768.0 tensor |
2 | $4.61 | 18.480 | Launch | ||
32,768.0 |
1 | $4.74 | 15.948 | Launch | ||
32,768.0 tensor |
2 | $9.40 | 36.306 | Launch | ||
Contact our dedicated neural networks support team at support@immers.cloud or send your request to the sales department at sale@immers.cloud.