Phi 3 Small 8K Instruct
May 2024
Phi 3 Small 8K InstructCompare inon
Benchmarks of Phi 3 Small 8K Instruct14
BenchmarkByScoreRank
No benchmark matches
—per task?
self-host onlyprice
—output speed?
8Kcontext
Ranking?
Alternatives?
ModelScoreCostPer taskWhat changes
BetterGPT-5 Minihighwins 75%—$0.053
OpenGemma 3 12Bunstatedwins 76%—$0.05 / $0.10Memory (4-bit): 3.7 GB → 6.1 GBBetterGPT-6 Lunamaxwins 83%—$0.068Function calling only with none · Chat Completions+3BetterGPT-5.6 Lunaxhighwins 83%—$0.085BetterGPT-5.6 Terraxhighwins 85%—$0.63Prompts with >272K input tokensModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
GPT-5 Minihighwins 75%—$0.053GPT-6 Lunamaxwins 83%—$0.068Function calling only with none · Chat Completions+3GPT-5.6 Lunaxhighwins 83%—$0.085GPT-5.6 Terraxhighwins 85%—$0.63Prompts with >272K input tokensGPT-6.1 Solmaxwins 87%—$0.72Remove unsupported parameters+5ModelScoreCostPer taskWhat changes
Gemma 3 12BunstatedOpenwins 76%—$0.05 / $0.10Memory (4-bit): 3.7 GB → 6.1 GBGemma 3n E4B InstructunstatedOpenwins 72%——Memory (4-bit): 3.7 GB → 3.9 GB
NVIDIA Nemotron 3.5 Lightning 30B A3B NVFP4Openwins 70%——Memory (4-bit): 3.7 GB → 11 GBGemma 3 4BunstatedOpenwins 69%—$0.04 / $0.05
Granite 4.1 8BunstatedOpenwins 68%——Memory (4-bit): 3.7 GB → 4.4 GBModelScoreCostPer taskWhat changes
Phi 4unstatedOpenwins 61%—$0.056 / $0.11Memory (4-bit): 3.7 GB → 7.3 GBPhi 3 Medium 4K InstructOpenwins 52%—$0.14 / $0.14Context: 8K → 4K+1Phi 3 Mini 128K InstructOpenwins 43%——Phi 3 Mini 4K InstructOpenwins 41%—$0.40 / $0.80Context: 8K → 4KPhi 4 Reasoning PlusOpenn/a—$0.07 / $0.35Memory (4-bit): 3.7 GB → 7.3 GBPhi 4 ReasoningOpenn/a——Memory (4-bit): 3.7 GB → 7.3 GBPhi 3.5 Mini InstructOpenn/a——Phi 3 Medium 128K InstructOpenn/a——Memory (4-bit): 3.7 GB → 7 GBAll Microsoft models →Run it yourself
estimateFits on✓ 8 GB (phone, laptop) · 8-bit✓ 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 · overall1,171#187
Arena Text · Chinese1,060.7#173
Arena Text · English1,150.5#179
Arena Text · French1,134.8#143
Arena Text · German1,079.7#147
Arena Text · instruction following1,086.5#179
Arena Text · Japanese965.6#134
Arena Text · Korean893.9#143
Arena Text · non-English1,057.9#179
Arena Text · Russian1,110.5#175
Arena Text · Spanish1,111.1#149
Arena Text · overall · full cohort1,109.8#178
Writing & Language
Arena Text · creative writing1,133#187
Arena Text · creative writing · full cohort1,082.7#179
FAQ
How much does Phi 3 Small 8K Instruct cost?›
There is no per-unit API price: self-host only.
Is Phi 3 Small 8K Instruct free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is Phi 3 Small 8K Instruct, and what is it best at?›
#187 of 196 in Chat & Knowledge, #187 of 195 in Writing & Language.
Is Phi 3 Small 8K Instruct better than Llama 3.2 3B Instruct?›
On Arena Text · creative writing, Phi 3 Small 8K Instruct scores 1,133 vs 1,145 — about the same.
What is the context window of Phi 3 Small 8K Instruct?›
8K tokens.
When was Phi 3 Small 8K Instruct released? Is there a newer version?›
Released 2024-05-07. It is the current version in its line.
Is Phi 3 Small 8K Instruct open source?›
Yes, the weights are open (MIT licence).
Can I run Phi 3 Small 8K Instruct locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Phi 3 Small 8K Instruct read images?›
No, it accepts text only.
What is Phi 3 Small 8K Instruct?›
Phi 3 Small 8K Instruct is an open-weights language model from Microsoft, released May 2024.
What are the best alternatives to Phi 3 Small 8K Instruct?›
GPT-5 Mini (better), Gemma 3 12B (open weights).