Phi-4-Reasoning-Vision-15B
No published training cutoff
Price not published
Small open model for local/edge reasoning, math and English text tasks.
Model idPhi-4
Price$0.07 in, $0.14 out per 1M tokens
Checked29 September 2026
At least2,045events since its release on 12 December 2024
93 historic, 480 major. 701 days of news.
Microsoft doesn’t publish a training cutoff, so we count from its release date. The real gap is larger. Coverage before 2026 is thin (34 entries for 2024 and 80 for 2025), so this count is a floor.
3 ways to use Phi-4 (14B), with the model id each provider expects.
| Provider | Model id | Endpoint | Docs |
|---|---|---|---|
| Hugging Face | huggingface.co | ||
| Azure AI Foundry | Phi-4 |
azure.microsoft.com | |
| OpenRouter | microsoft/phi-4 |
openrouter.ai |
| Input | $0.07 per 1M tokens |
|---|---|
| Output | $0.14 per 1M tokens |
As published: “per 1M tokens (USD) on OpenRouter; self-hosting free”.
Source: openrouter.ai, checked 29 September 2026.
What its maker marketed at launch, and what people found later, each with its source.
14B dense model trained on 9.8T tokens heavy in curated synthetic data, prioritizing reasoning over scale (84.8 MMLU, 80.4 MATH).
Found after launch
Base for Phi-4-reasoning, Phi-4-reasoning-plus, Phi-4-mini(-reasoning/-flash-reasoning) and Phi-4-multimodal-instruct open models.
No Phi-5 found on Hugging Face as of 2026-09-29 (microsoft org). Siblings: microsoft/Phi-4-mini-instruct, microsoft/Phi-4-reasoning-plus, microsoft/Phi-4-multimodal-instruct.
from transformers import pipeline
pipe = pipeline("text-generation", model="microsoft/phi-4", device_map="auto")
print(pipe([{"role":"user","content":"Solve 2x+3=11"}], max_new_tokens=128))
Sources: https://huggingface.co/microsoft/phi-4 , https://azure.microsoft.com/en-us/products/phi
Not yet checked against an official source: Foundry model name not re-verified today.
Paste this into a new chat. It points your AI to a dated list of AI news since October 2024, with a source for every item.
Using Gemini, Grok or another assistant? Copy the prompt and paste it into a new chat. Only Claude and ChatGPT have a link that opens with a prompt filled in.
Files for every cutoff, with their sizes: Briefings.
For developers: an MCP server is planned. For AI agents