References 589% CONFIDENCEOverall confidence: 89%How well the pin's source and references back up its dates.Weighted average of how firmly 5 references, the source included, support the pin's start and end times; a reference counts half as much for every 180 days older than the newestShow all pins at 75% confidence or better

The first entry is always the pin's source. Overall confidence is a weighted average of how firmly each reference supports the start and end times used above; a reference counts half as much for every 180 days older than the newest.

• [1]90%
  techcommunity.microsoft.com/blog/azure-ai-foundry-blog/introducing-phi-4-microsoft%E2%80%99s-newest-small-language-model-specializing-in-comple/4357090

  techcommunity.microsoft.com· Posted Oct 12, 2026· Starts Dec 12, 2024 ✓· 76% of score

  Microsoft's Azure AI Foundry blog post "Introducing Phi-4" is dated Dec 12, 2024, the Hugging Face^[4] model card gives a release date of December 12, 2024 and the technical report was submitted to arXiv^[5] the same day; the open weights followed on January 8, 2025 and OpenRouter^[2] lists it from Jan 10, 2025.

• [2]80%
  Microsoft: Phi 4 - OpenRouter

  openrouter.ai· Published Jan 10, 2025· 6% of score

  Router listing: released Jan 10, 2025, 16K context and $0.07 per 1 million input tokens and $0.14 per 1 million output tokens.^[1]

• [3]80%
  Microsoft releases full Phi-4 model with weights under MIT license - The Decoder

  the-decoder.com· Published Jan 8, 2025· 6% of score

  The Decoder's December 13 report plus its January 8, 2025 update that the full weights are on Hugging Face^[4] under the MIT licence; adds 56.1% GPQA, 80.4% MATH, about 400B synthetic tokens in 50 dataset types, and the weakness at strict instruction following.

• [4]90%
  microsoft/phi-4 - Hugging Face

  huggingface.co· Published Dec 12, 2024· 6% of score

  Model card: 14B dense decoder-only, 16K context, MIT licence, release date December 12, 2024, 9.8T training tokens in 21 days on 1,920 H100-80G GPUs, June 2024 data cut-off, mostly English, and the SimpleEval table against GPT-4o and Llama-3.3.^[1]

• [5]90%
  Phi-4 Technical Report - arXiv

  arxiv.org· Published Dec 12, 2024· 6% of score

  Microsoft^[1] Research's report, submitted 12 December 2024: synthetic data throughout pre-training, post-training innovations, and STEM question answering that "substantially surpasses its teacher model" with only a minimally changed phi-3 architecture.

Suggest a correction

Something missing or wrong? Say it in your own words: a link that backs this pin up, a different start or end date and why, or a fact it lacks or gets wrong. The AI checks it against this pin's sources, searches for better ones, and adds any page that backs you up. The pin's own sources still count most. A picture that shows something else, or shows it badly, is looked at too, and moved down or replaced.