Inference
EN
OpenBMB

MiniCPM5-1B

MiniCPM5-1B is a reasoning model from OpenBMB in the MiniCPM5 family. 13 benchmarks count toward its index, in 6 categories.

  • Open weights
  • Reasoning
  • Input: text
  • Released 2026-05
RankedEnough results for an index.
IndexAll categories, out of 100.15.5±7.0Median 51.0
CoverageIndex weight that has results.75%
SpeedOutput tokens per second.—Median 67/s
First tokenSeconds to the first output token.—Median 1.3 s
Input / 1MUS dollars per 1M input tokens.—Median $0.60
Output / 1MUS dollars per 1M output tokens.—Median $2.55
ContextMaximum tokens in one request.131KMedian 262K
EloLMArena rating and rank.—Median 1422

Run locally

1.08B parameters
16-bit
3.3 GB
8-bit
2.2 GB
4-bit
1.7 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy fits in 8 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.1 counted8.5—
CodingCode writing and repair.1 counted11.9—
ReasoningLogic problems and puzzles.None counted——
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.4 counted9.2268 / 268
MultilingualTasks in many languages.1 counted5.0—
InstructionTasks with strict prompt rules.2 counted28.954 / 63
MathMath problems.4 counted21.9159 / 168

Results

13 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
MATH-500 Problem SetMathAbove fitFar above this model's index.91.6%19 / 27
Instruction-Following EvalInstruction80.4%54 / 70
American Invitational Mathematics Examination 2025Math40.4%46 / 53
Massive Multitask Language Understanding ProfessionalKnowledge48.9%66 / 75
Instruction Following BenchmarkInstructionAbove fitFar above this model's index.46.7%52 / 57
Berkeley Function Calling Leaderboard v4Agentic25.1%24 / 26
EuroEval FrenchMultilingual11.7% ±0.9166 / 178
EuroEval SpanishMultilingual9.7% ±0.8165 / 176
LiveCodeBench v6Coding33.5%57 / 59
EuroEval GermanMultilingual7.1% ±0.9185 / 191
AIME 2026Math40.4%33 / 34
GPQA Diamond · BenchLMKnowledge26.3%83 / 84
MMLU-ReduxKnowledge70.1%19 / 19
Harvard-MIT Mathematics Tournament February 2026Math25.8%27 / 27
SuperGPQA: Scaling LLM Evaluation Across 285 Graduate DisciplinesKnowledge23.1%22 / 22
Artificial Analysis Openness IndexKnowledge83.3%—
BIG-Bench HardReasoning71.9%—
LiveCodeBench ProCoding22.7%—

Sources

BenchLMCC BY-NC 4.0EuroEvalMIT

History

5 releases
Index16.1 → 15.5
23 Oct 202612 Nov 2026

Questions

How good is MiniCPM5-1B?
Its index is 15.5 out of 100. That is place 281 of 283 ranked models.
What is MiniCPM5-1B best at?
Instruction: 28.9 out of 100, place 54 of 63.
What is the context window of MiniCPM5-1B?
131K tokens.
Is MiniCPM5-1B open source?
Yes. Its weights are published.
When was MiniCPM5-1B released?
May 2026.

Badge

MiniCPM5-1B on Inference360: place 281 of 283, index 15.5
HTML
<a href="https://inference360.ai/models/minicpm5-1b"><img src="https://inference360.ai/badges/minicpm5-1b.svg" alt="MiniCPM5-1B on Inference360: place 281 of 283, index 15.5" height="20"></a>
Markdown
[![MiniCPM5-1B on Inference360: place 281 of 283, index 15.5](https://inference360.ai/badges/minicpm5-1b.svg)](https://inference360.ai/models/minicpm5-1b)

Compare with

Granite-4.0-H-1B16.3Llama 3.1 8B17.9Granite-4.0-H-350M12.7Granite-4.0-350M11.9MiniCPM5-2B37.1

More from OpenBMB

All models
MiniCPM5-2B37.1MiniCPM-o 2.6Unranked