Loading...
Loading...
[Microsoft Research](/microsoft) Phi-4 is designed to perform well in complex reasoning tasks and can operate efficiently in situations with limited memory or where quick responses are needed. At 14 billion parameters, it was trained on a mix of high-quality synthetic datasets, data from curated websites, and academic materials. It has undergone careful improvement to follow instructions accurately and maintain strong safety standards. It works best with English language inputs. For more information, please see [Phi-4 Technical Report](https://arxiv.org/pdf/2412.08905)
Quality Index
10.4
355th of 442
Top 80%
Coding Index
11.2
244th of 352
Top 70%
Math Index
18.0
213th of 268
Top 80%
Price/1M
$0.22
303rd cheapest
29% below median
Top 45%
Speed
27 tok/s
Top 60%
TTFT
0.43s
Context Window
16K
326th largest
Top 93%
Input
$0.13
per 1M tokens
Output
$0.50
per 1M tokens
Blended
$0.22
per 1M tokens
Cheaper than 55% of models. Median price is $0.31/1M tokens.
Daily
$0.22
Monthly
$6.57
27
tokens/sec
Faster than 40% of models
0.43
seconds
Faster than 49% of models
0.43
seconds
Faster than 51% of models
Market Median
46 tok/s
42% slower
Median TTFT
0.42s
3% slower
Throughput/Dollar
121
tok/s per $/1M
Speed Comparison
Context Window
16K
tokens
Larger than 7% of models
Max Output
16K
tokens
100% of context