MMODELYST
Models/Phi-4 Multimodal Instruct
MDL

Phi-4 Multimodal Instruct

Microsoft

Verified against Artificial Analysis · Jul 13, 2026
Capability
42
percentile index
Price /M
$0 / $0
in / out
Context
Avg score
29.6
6 benchmarks
AA Intelligence
4.5
index
Speed
17 tok/s
TTFT 0.4s
Released
Feb 26, 2025
About

Phi-4 Multimodal Instruct is a open-weights model in the Phi family from Microsoft. Benchmarked on 6 evals, averaging 29.6.

Benchmark scores · 6
avg 29.6 — higher is better
MATH
69.3
MMLU-Pro
48.5
GPQA Diamond
31.5
LiveCodeBench
13.1
SciCode
11
Humanity's Last Exam
4.4
The research behind this model
all papers mentioning it →

Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.

Details
PublisherMicrosoft
FamilyPhi
Parameters
License
WeightsOpen
ReleasedFeb 26, 2025
Performance & price
Output speed17 tok/s
Latency (TTFT)0.4s
Price in / out$0 / $0 per 1M

Median across API providers, via Artificial Analysis. Blended = 3:1 input:output. For reasoning models, latency includes thinking time. Methodology →

Capability over time

Append-only ledger — every observed change to this model's numbers.

API providers
Artificial Analysis$0/M