phi-4Microsoft
- endpoints
- 2
- regions
- 1
- api
- chat
- released
- January 2025
capabilities 1/8
The same weights cost 1.3x more on deepinfra inc. than on deepinfra inc.. That is what the endpoints table is for.
| # | flags | ||||||
|---|---|---|---|---|---|---|---|
| 1 |  DeepInfra Inc. | Global | 16K | $0.06 | $0.11 | - | zdr |
| 2 | DeepInfra Inc. | Global | 16K | $0.07 | $0.14 | $0.07 | zdr |
no measured traffic for this model yet. the endpoints table above carries provider pricing
| eval | Score as a share of 100 | /100 |
|---|---|---|
| MMLU Pro | 71.4 | |
| GPQA Diamond | 57.5 | |
| LiveCodeBench | 23.1 | |
| Math Index | 18.0 | |
| AIME 2025 | 18.0 | |
| Intelligence Index | 5.9 | |
| Terminal-Bench Hard | 3.8 | |
| Humanity's Last Exam | 3.8 | |
| τ²-Bench | 0.0 | |
| 9 evals | mean | 22.4 |
Scores from Artificial Analysis and public leaderboards, normalised to 100. Bar colour is the band, not the rank: green 80 and up, blue 55 and up, amber 30 and up. Benchmarks measure narrow skills, so test on your own workload before committing. Released 2024-12-12.
deepinfra/microsoft/phi-4:flexthis model has no managed policy yet, so call the provider endpoint directly. every id in the endpoints table works the same way
Base url is https://router.requesty.ai/v1 for every id here. One key reaches the whole catalog.
from openai import OpenAI client = OpenAI( api_key="YOUR_REQUESTY_API_KEY", base_url="https://router.requesty.ai/v1",) response = client.chat.completions.create( model="deepinfra/microsoft/phi-4:flex", messages=[ {"role": "user", "content": "Explain quantum computing in one paragraph."}, ],) print(response.choices[0].message.content)
Change the base url, use your Requesty key, set the model to any id in the call pane. Existing OpenAI SDK code needs no other edit, and the same key reaches every other model in the catalog, so switching later is a one-parameter change. Browse all models
Phi-4-reasoning-plus is an enhanced 14B parameter model from Microsoft, fine-tuned from Phi-4 with additional reinforcement learning to boost accuracy on math, science, and code reasoning tasks. It uses the same dense decoder-only transformer architecture as Phi-4, but generates longer, more comprehensive outputs structured into a step-by-step reasoning trace and final answer. While it offers improved benchmark scores over Phi-4-reasoning across tasks like AIME, OmniMath, and HumanEvalPlus, its responses are typically ~50% longer, resulting in higher latency. Designed for English-only applications, it is well-suited for structured reasoning workflows where output quality takes priority over response speed.
questions 5
Which providers serve phi-4?
How much does phi-4 cost?
Is phi-4 open weights?
What is the context window of phi-4?
How do I use phi-4 with the OpenAI SDK?
more from microsoft 1
| model | context | from /M | providers |
|---|---|---|---|
| wizardlm-2-8x22b | 66K | $0.62 | 1 |
route phi-4 through one endpoint
One key for 1 provider on this model and 600+ others. No markup on provider prices, automatic failover, caching built in. Weekly aggregates in the measured pane come from production traffic routed through Requesty, one line per provider on a shared axis. Methodology