MDL
phi-4
microsoft
Verified against Artificial Analysis · Jul 13, 2026
Capability
42
percentile index
Price /M
$0.125 / $0.5
in / out
Context
16.4K
tokens
Avg score
27.8
15 benchmarks
AA Intelligence
4.9
index
Speed
31 tok/s
TTFT 0.6s
Released
Dec 12, 2024
About
phi-4 is a open-weights model in the phi family from microsoft. Benchmarked on 15 evals, averaging 27.8.
mittext-generation
Benchmark scores · 15
avg 27.8 — higher is betterThe research behind this model
all papers mentioning it →Phi-4 Technical ReportPhi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs4 cites · Mar 3, 2025X-Token: Projection-Guided Cross-Tokenizer Knowledge DistillationMay 20, 2026LamPO: A Lambda Style Policy Optimization for Reasoning Language ModelsMay 20, 2026GraphRAG on Consumer Hardware: Benchmarking Local LLMs for Healthcare EHR Schema RetrievalMay 20, 2026Fine-tuning with Hierarchical Prompting for Robust Propaganda Classification Across Annotation SchemasMay 13, 2026
Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.
Performance & price
Output speed31 tok/s
Latency (TTFT)0.6s
Price in / out$0.125 / $0.5 per 1M
Blended price$0.219 / 1M
Math index18.0
Median across API providers, via Artificial Analysis. Blended = 3:1 input:output. For reasoning models, latency includes thinking time. Methodology →
History
full ledger →Capability over time
Append-only ledger — every observed change to this model's numbers.
API providers
Artificial Analysis$0.5/M