Phi-4 vs Phi-4-reasoning
Change modelsPhi-4Microsoft
Phi-4-reasoningMicrosoft
IndexAll categories, out of 100.
22.4 ±5.7
27.1 ±6.4 best
GapIndex points below the top.
−4.7 · no clear gap
Top
CoverageIndex weight that has results.
90% best
85%
SpeedOutput tokens per second.
93/s best
41/s
EloLMArena rating and rank.
1217
—
ReleasedMonth of release.
2024-12
2025-04
Input / 1MUS dollars per 1M input tokens.
$0.07
—
Output / 1MUS dollars per 1M output tokens.
$0.14
—
1K chats2K tokens in, 500 out, each.
$0.21
—
1K documents20K tokens in, 1K out, each.
$1.54
—
1K agent tasks200K tokens in, 20K out, each.
$16.80
—
ContextMaximum tokens in one request.
16K
32K best
AgenticMulti-step tasks with tools.
5.0
5.0
CodingCode writing and repair.
23.4
34.0 best
ReasoningLogic problems and puzzles.
22.1
22.1
MultimodalTasks with images and text.
—
—
KnowledgeFacts and expert knowledge.
29.9
33.3 best
MultilingualTasks in many languages.
24.0
—
InstructionTasks with strict prompt rules.
11.0
21.1 best
MathMath problems.
23.7
35.5 best
Benchmarks
7 sharedOnly Phi-4: 7Only Phi-4-reasoning: 5
Phi-4
Phi-4-reasoning