Phi-3 Mini 128K

Released
2024-04-23
Last refreshed
2026-10-07
Status
Researched 60d ago
Open sourceCommercial use: permittedCodingRAGLong contextClassification
Microsoft Research releases · 7 in the last 12 monthsChangelog →
Specifications
Family
Phi-3
Released
2024-04-23
Context
128K
Parameters
3.8B
Architecture
Decoder Only
Knowledge cutoff
2023-10
Specialization
general
Openness
Open source
License
MITOSI-approvedPermitted
Weights
Unknown
Code
Unknown
Training
Fine-tuned
Created by

Advancing the state-of-the-art in AI and computing.

Redmond, Washington, United States
Founded 1991
Website
Pricing
Output / 1M
$0.100
Input / 1M
$0.100

Cheapest of 3 routes · Fireworks AI

About

Phi-3 Mini-128K-Instruct, developed by Microsoft, is a 3.8 billion-parameter large language model renowned for its lightweight, open-source architecture. Despite its modest size, it excels in reasoning tasks, particularly in math and logic, and showcases strong code generation capabilities. A standout feature is its remarkable ability to handle up to 128,000 tokens, allowing it to process extensive text documents and codebases efficiently. While it has limitations in factual knowledge and focuses primarily on English, it strikes a balance between performance and efficiency, making it ideal for resource-constrained environments.

Providers(3)

Compare all 3

Compare API pricing across 3 tracked provider routes for input and output tokens, batch, and cached reads when available.

ProviderInput / 1MOutput / 1MRoute
Fireworks AI$0.100$0.100
Provisioned
Replicate API$0.050$0.250
Serverless
Baseten API--
ServerlessPartial

Deprecated provider routes(2)

ProviderStatusCatalog ID
NVIDIA NIM
DeprecatedRetired April 15, 2026
—
Microsoft Foundry
DeprecatedRetired August 30, 2025
—

Available via routers & gateways(7)

Capabilities

No model capability flags are currently sourced.

Benchmark peer barsfor Coding

Benchmark scores(9)

Scores are benchmark-specific and are direction-aware: the same numeric gap can mean very different outcomes across suites. Use the leaderboard context and this model's provider route to decide whether the winning margin is meaningful for your workload.
BenchmarkScoreVersionEvaluationSource
50.8diamondObserved 2026-03-06—
90.210-shotObserved 2026-03-06—
75.9pass@1Observed 2026-03-06—
76.55-shotObserved 2026-03-06—
43.9—Observed 2026-04-14—
84.9—Observed 2026-05-28—
78.6—Observed 2026-05-28—
66.7RULER (128K) (accuracy)Observed 2026-10-26—
1024.0text creative_writing 2026-10-22Observed 2026-10-22
Configuration: Phi-3 Mini 128K Instruct
Harness: LMArena blind pairwise human votes, Bradley-Terry
Evaluator: LMArena
Confidence: confirmed
Notes: LMArena Text arena, Creative Writing category (no style control), board published 2026-10-22 (HF lmarena-ai/leaderboard-dataset @a467e6b9, text config, latest split). Board entry 'Phi-3 Mini 128K Instruct' (category creative_writing in text/latest-00000-of-00001.parquet): rank 396 of 416, rating 1024.0 (95% CI 1010.0-1037.9), 2949 votes. Source: LMArena (CC BY 4.0). Retrieved 2026-10-27.
Source

Source: LMArena (lmarena.ai), CC BY 4.0

Migration checks

No linked migration route is available for this model yet.