Phi 3 Mini 4K Instruct
Apr 2024
Phi 3 Mini 4K InstructCompare inon
Benchmarks of Phi 3 Mini 4K Instruct16
BenchmarkByScoreRank
No benchmark matches
Ranking?
Alternatives?
ModelScoreCostPer taskWhat changes
CheaperMiMo v2.6 FlashunstatedOpenwins 89%-65%$0.14 / $0.28Memory (4-bit): 1.9 GB → 186 GB+4
BetterGPT-5 Minihighwins 82%+38%$0.053
OpenGemma 3 12Bunstatedwins 79%-88%$0.05 / $0.10Memory (4-bit): 1.9 GB → 6.1 GB
CheaperGLM 5.3 FlashunstatedOpenwins 89%-53%$0.15 / $0.50Memory (4-bit): 1.9 GB → 193 GB+5CheaperGemma 4 31BunstatedOpenwins 87%-68%$0.10 / $0.34Memory (4-bit): 1.9 GB → 19 GBModelScoreCostPer taskWhat changes
MiMo v2.6 FlashunstatedOpenwins 89%-65%$0.14 / $0.28Memory (4-bit): 1.9 GB → 186 GB+4GLM 5.3 FlashunstatedOpenwins 89%-53%$0.15 / $0.50Memory (4-bit): 1.9 GB → 193 GB+5Gemma 4 31BunstatedOpenwins 87%-68%$0.10 / $0.34Memory (4-bit): 1.9 GB → 19 GB
Hy3unstatedOpenwins 87%-54%$0.13 / $0.53Memory (4-bit): 1.9 GB → 179 GBMiMo v2.5unstatedOpenwins 87%-65%$0.14 / $0.28Memory (4-bit): 1.9 GB → 186 GBModelScoreCostPer taskWhat changes
GPT-5 Minihighwins 82%+38%$0.053GPT-6 Lunamaxwins 83%-60%$0.068Function calling only with none · Chat Completions+3
Claude Haiku 5.5highwins 84%-60%$0.079Remove sampling parameters+5GPT-5.6 Lunaxhighwins 87%-10%$0.085GPT-5.6 Terraxhighwins 87%+800%$0.63Prompts with >272K input tokensModelScoreCostPer taskWhat changes
Gemma 3 12BunstatedOpenwins 79%-88%$0.05 / $0.10Memory (4-bit): 1.9 GB → 6.1 GB
Granite 4.2 8BunstatedOpenwins 78%-79%$0.06 / $0.25Memory (4-bit): 1.9 GB → 4.4 GBGranite 4.2 3BunstatedOpenwins 76%-90%$0.03 / $0.12Gemma 3n E4B InstructunstatedOpenwins 76%——Memory (4-bit): 1.9 GB → 3.9 GBGranite 4.1 8BunstatedOpenwins 76%——Memory (4-bit): 1.9 GB → 4.4 GBModelScoreCostPer taskWhat changes
Phi 4unstatedOpenwins 66%—$0.056 / $0.11Memory (4-bit): 1.9 GB → 7.3 GBPhi 3 Medium 4K InstructOpenwins 57%—$0.14 / $0.14Memory (4-bit): 1.9 GB → 7 GBPhi 3 Small 8K InstructOpenwins 55%——Memory (4-bit): 1.9 GB → 3.7 GBPhi 3 Mini 128K InstructOpenwins 45%——Phi 4 Reasoning PlusOpenn/a—$0.07 / $0.35Memory (4-bit): 1.9 GB → 7.3 GBPhi 4 ReasoningOpenn/a——Memory (4-bit): 1.9 GB → 7.3 GBPhi 3.5 Mini InstructOpenn/a——Phi 3 Medium 128K InstructOpenn/a——Memory (4-bit): 1.9 GB → 7 GBAll Microsoft models →Run it yourself
estimateFits on✓ 8 GB (phone, laptop) · full quality✓ 16 GB · full quality✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
LicenceMITfree for commercial use
Benchmarks
all categoriesChat & Knowledge
Arena Text · overall1,128#192
Arena Text · Chinese1,021.1#193
Arena Text · English1,117.2#200
Arena Text · French1,076.3#160
Arena Text · German1,043.9#163
Arena Text · instruction following1,053.2#200
Arena Text · Japanese935#149
Arena Text · Korean905.4#154
Arena Text · non-English1,020.6#198
Arena Text · Russian1,022.1#197
Arena Text · Spanish1,084.8#165
Arena Text · overall · October 22, 20261,073.3#193
Arena Text · overall · October 15, 20261,073.3#183
Writing & Language
Arena Text · creative writing1,073#194
FAQ
How much does Phi 3 Mini 4K Instruct cost?›
$0.40 per 1M input tokens and $0.80 per 1M output tokens (cheapest host: Featherless).
Is Phi 3 Mini 4K Instruct free to use?›
There is no free API; the cheapest paid price is $0.40 per 1M input tokens. The weights are open, so you can run it yourself.
How good is Phi 3 Mini 4K Instruct, and what is it best at?›
#200 of 205 in Chat & Knowledge, #192 of 197 in Writing & Language.
Is Phi 3 Mini 4K Instruct better than Gemma 1.1 7B It?›
On Arena Text · English, Phi 3 Mini 4K Instruct scores 1,117.2 vs 1,124 — about the same.
How can I use Phi 3 Mini 4K Instruct?›
Through the Microsoft API.
What is the context window of Phi 3 Mini 4K Instruct?›
4K tokens.
When was Phi 3 Mini 4K Instruct released? Is there a newer version?›
Released 2024-04-22. It is the current version in its line.
Is Phi 3 Mini 4K Instruct open source?›
Yes, the weights are open (MIT licence).
Can I run Phi 3 Mini 4K Instruct locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Phi 3 Mini 4K Instruct read images?›
No, it accepts text only.
What is Phi 3 Mini 4K Instruct?›
Phi 3 Mini 4K Instruct is an open-weights language model from Microsoft, released Apr 2024.
What are the best alternatives to Phi 3 Mini 4K Instruct?›
MiMo v2.6 Flash (cheaper), GPT-5 Mini (better), Gemma 3 12B (open weights).