Microsoft model product
Phi 3.5 mini
8B-parameter language model from Microsoft with a 128K-token context window and support for over 20 languages.
Updated Sep 8, 2026. Default version: Phi-3.5-mini-instruct
LLMBoard Score0.0Phi-3.5-mini-instruct
Coverage60%21 benchmark families
Context window128KTokens
Official input priceN/AOfficial price unavailable
Phi 3.5 mini Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Phi 3.5 mini Benchmark Results
Benchmark scores for Phi-3.5-mini-instruct.
Sort by
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkQasper | Score41.90% | Rank01 | Participants2 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkQMSum | Score21.30% | Rank01 | Participants2 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkSQuALITY | Score24.30% | Rank01 | Participants5 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkGovReport | Score25.90% | Rank02 | Participants2 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkMEGA MLQA | Score61.70% | Rank02 | Participants2 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkMEGA TyDi QA | Score62.20% | Rank02 | Participants2 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkMEGA UDPOS | Score46.50% | Rank02 | Participants2 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkMEGA XCOPA | Score63.10% | Rank02 | Participants2 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkMEGA XStoryCloze | Score73.50% | Rank02 | Participants2 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkOpenBookQA | Score79.20% | Rank02 | Participants5 | Percentile75.00% | EvidenceC | Evaluated |
| BenchmarkRepoQA | Score77.00% | Rank02 | Participants2 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkSocial IQa | Score74.70% | Rank02 | Participants9 | Percentile87.50% | EvidenceC | Evaluated |
| BenchmarkSummScreenFD | Score16.00% | Rank02 | Participants2 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkRULER | Score84.10% | Rank04 | Participants4 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkTruthfulQA | Score64.00% | Rank05 | Participants18 | Percentile76.47% | EvidenceC | Evaluated |
| BenchmarkPIQA | Score81.00% | Rank07 | Participants11 | Percentile40.00% | EvidenceC | Evaluated |
| BenchmarkBoolQ | Score78.00% | Rank08 | Participants10 | Percentile22.22% | EvidenceC | Evaluated |
| BenchmarkARC-C | Score84.60% | Rank14 | Participants34 | Percentile60.61% | EvidenceC | Evaluated |
| BenchmarkBIG-Bench Hard | Score69.00% | Rank15 | Participants21 | Percentile30.00% | EvidenceC | Evaluated |
| BenchmarkWinogrande | Score68.50% | Rank18 | Participants22 | Percentile19.05% | EvidenceC | Evaluated |
| BenchmarkArena Hard | Score37.00% | Rank23 | Participants26 | Percentile12.00% | EvidenceC | Evaluated |
| BenchmarkMBPP | Score69.60% | Rank23 | Participants33 | Percentile31.25% | EvidenceC | Evaluated |
| BenchmarkHellaSwag | Score69.40% | Rank25 | Participants27 | Percentile7.69% | EvidenceC | Evaluated |
| BenchmarkMGSM | Score47.90% | Rank31 | Participants31 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkGSM8k | Score86.20% | Rank35 | Participants48 | Percentile27.66% | EvidenceC | Evaluated |
| BenchmarkMMMLU | Score55.40% | Rank48 | Participants49 | Percentile2.08% | EvidenceC | Evaluated |
| BenchmarkMATH | Score48.50% | Rank57 | Participants71 | Percentile20.00% | EvidenceC | Evaluated |
| BenchmarkHumanEval | Score62.80% | Rank62 | Participants66 | Percentile6.15% | EvidenceC | Evaluated |
| BenchmarkMMLU | Score69.00% | Rank87 | Participants101 | Percentile14.00% | EvidenceC | Evaluated |
| BenchmarkMMLU-Pro | Score47.40% | Rank125 | Participants138 | Percentile9.49% | EvidenceC | Evaluated |
Phi 3.5 mini Arena Results
Preference and agent-evaluation results for the default version.
No Arena results
The default version does not have a matching Arena result yet.
Phi 3.5 mini Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- N/A
- Official provider
- N/A
- Lowest third-party
- N/A
- Tracked offerings
- 0
No provider prices
The default version has no current input or output token prices.
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Phi 3.5 mini Runtime Performance
Provider-specific output speed and catalog latency for Phi-3.5-mini-instruct. Runtime does not affect the capability score.
Sort by
Provider | Output Speed | Catalog Latency | Max Input | Max Output | Updated |
|---|
| ProviderAzure | Output Speed23.00 tok/s | Catalog Latency0.52 s | Max Input128K | Max Output128K | Updated |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Phi 3.5 mini Specifications
Technical details for the model's default version.
- Version
- Phi-3.5-mini-instruct
- Released
- Aug 23, 2024
- Knowledge cutoff
- Unknown
- Parameters
- 3.8B
- Context window
- 128K
- Max output
- 128K
- Inputs
- text
- Outputs
- text
- Open weights
- Yes
- License
- MIT
Phi 3.5 mini Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
Sort by
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionPhi-3.5-mini-instruct | Released | LLMBoard0.00 | Parameters3.8B | Context128K | Max output128K | Open weightsYes | LicenseMIT |
Models similar to Phi 3.5 mini
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
MI
Phi 3.5 vision
Microsoft
MI
Phi 4 Mini
Microsoft
MI
Phi 3.5 MoE
Microsoft
MI
Phi 4 multimodal
Microsoft
MI
Phi 4
Microsoft
MI
Phi 4 Mini Reasoning
Microsoft
What is Phi 3.5 mini?
Key information about Phi 3.5 mini and its available data.
8B-parameter model from Microsoft that supports up to 128K context tokens and multilingual use across over 20 languages. It was additionally trained and safety post-trained for instruction-following, reasoning, math, and code generation, and is released under the MIT license.
Data as of 2026-09-08.
FAQ
Common questions about Phi 3.5 mini.
When was Phi 3.5 mini released?
Phi 3.5 mini's default version was released on Aug 23, 2024.
How much does Phi 3.5 mini cost?
No official standard PAYG price is currently available for Phi 3.5 mini.
Who created Phi 3.5 mini?
Phi 3.5 mini was created by Microsoft.
What is the context window for Phi 3.5 mini?
The default version has a 128K token context window.
Is Phi 3.5 mini open weight?
Yes. The default version is marked as open weight under MIT.
How many API providers offer Phi 3.5 mini?
No provider offering is currently linked to the default version.