MiniCPM5-2B
- Input
- —
- Output
- —
- Speed
- ~110tok/sest.
- Context
- 131Ktokens
Long-tail import from Artificial Analysis (2026-09-24); AA variant: MiniCPM5-2B. Price = AA median host price.
Rankings
Self-hostingEstimated
How we estimate →What it would cost to run these open weights yourself on rented GPUs. No API sells this model, so this is its price on the chart.
- Hardware
- 1× B200 180GB
- BF16 weights
- Throughput
- ~20,229 tok/s
- many requests batched
- Price
- $0.035–0.31 /M tok
- busy → light use
- On your own machine
- RTX 5090 32GB (int4, 300+ tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- BF16 weights (2.52B params) + 25% KV-cache headroom ≈ 6 GB
- 1× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
- ~20,229 output tok/s aggregate at batch 256 (bandwidth-bound)
- Blended 3:1 input:output; prefill ~210,938 tok/s
- Low = 75% utilization at the low GPU price; high = 20% at the high price
Benchmark scores
~ italic, dashed = no published score yet, estimated from related benchmarks.
Coding
SWE-bench VerifiedEstimated
~#18 of 21 (estimated)~62.7%
- #1Claude Opus 596.0%
- #2Claude Sonnet 585.2%
- #17Intern-S2-Preview (35B-A3B)64.0%
- ~#18MiniCPM5-2BEstimated62.7%
- #18gpt-oss-120b62.4%
No published score — estimated from related benchmarks
SWE-bench Pro (Public)Estimated
~#23 of 24 (estimated)~45.4%
- #1Claude Fable 5.293.0%
- #2Claude Opus 5.589.9%
- #22Muse Glimmer51.2%
- ~#23MiniCPM5-2BEstimated45.4%
- #23Nex-N2.5-Mini43.8%
No published score — estimated from related benchmarks
DeepSWE v1.1Estimated
~#43 of 46 (estimated)~33.2%
- #1Gemini 4 Argon78.8%
- #2DeepSeek V4.1 Flash 101777.0%
- #42Nex-N2.5-Mini36.1%
- ~#43MiniCPM5-2BEstimated33.2%
- #43Hy328.0%
No published score — estimated from related benchmarks
Terminal-Bench 4.0Verified
#63 of 1010.0%
- #1Kimi K3.1 Pro67.2%
- #2Claude Sonnet 5.563.6%
- #63MiMo-V2.50.0%
- #63MiniCPM5-2B0.0%
- #63Ministral 3 14B0.0%
artificialanalysis.ai · 2026-09-24
LiveCodeBenchEstimated
~#20 of 28 (estimated)~47.8%
- #1Qwen3.8-Omni-Flash92.6%
- #2Qwen3.8 Flash-Next91.9%
- #19Falcon-H1-34B-Instruct49.7%
- ~#20MiniCPM5-2BEstimated47.8%
- #20Mistral Large 346.5%
No published score — estimated from related benchmarks
SciCodeVerified
#107 of 11826.3%
- #1Claude Fable 5.267.7%
- #2Claude Opus 5.566.9%
- #106K2 Horizon 7B27.5%
- #107MiniCPM5-2B26.3%
- #108Granite 4.2 3B25.3%
artificialanalysis.ai · 2026-09-24
AA Coding Agent IndexEstimated
~#25 of 26 (estimated)~27.9
- #1Claude Opus 5.561.7
- #2Claude Opus 559.7
- #24GPT-6 Luna31.1
- ~#25MiniCPM5-2BEstimated27.9
- #25Muse Spark Open26.3
No published score — estimated from related benchmarks
Hard reasoning
GPQA DiamondVerified
#113 of 19270.2%
- #1GPT-6 Astra96.1%
- #2Gemini 3.8 Flash95.3%
- #112Qwen3.5 4B71.2%
- #113MiniCPM5-2B70.2%
- #114K2 Horizon 3.7B69.2%
artificialanalysis.ai · 2026-09-24
Humanity's Last ExamVerified
#126 of 1988.9%
- #1Claude Fable 5.274.3%
- #2Claude Opus 5.561.4%
- #125Ling 3.0 Tiny9.3%
- #126MiniCPM5-2B8.9%
- #127Ling-2.6-1T8.7%
artificialanalysis.ai · 2026-09-24
AIME (latest)Estimated
~#23 of 34 (estimated)~60.8%
- #1GLM-5.3 NVFP499.5%
- #2Beam 1.0 501B A23B97.7%
- #22Phi-4-reasoning-plus78.0%
- ~#23MiniCPM5-2BEstimated60.8%
- #23Mistral Large 338.0%
No published score — estimated from related benchmarks
Agents
τ²-bench (Telecom)Estimated
~#54 of 120 (estimated)~45.1%
- #1Step 3.7 Flash98.5%
- #2DeepSeek-V4-Pro-0813-nvfp4-DSpark98.3%
- #53Nemotron 3 Nano Omni 30B A3B Reasoning45.3%
- ~#54MiniCPM5-2BEstimated45.1%
- #54Gemma 4 26B A4B43.6%
No published score — estimated from related benchmarks
OSWorld-VerifiedEstimated
~#12 of 15 (estimated)~62.3%
- #1Kimi K384.8%
- #2Qwen3.8 27B84.3%
- #11Muse Glimmer65.9%
- ~#12MiniCPM5-2BEstimated62.3%
- #12Nemotron 3.5 Super VL NVFP452.3%
No published score — estimated from related benchmarks
OSWorld 2.0Estimated
~last of 5 (estimated)below measured range (<47.5%)
No published score — estimated from related benchmarks
BrowseCompEstimated
~#19 of 22 (estimated)~60.1%
- #1Atria Dawn Preview92.5%
- #2GPT-6 Astra91.5%
- #18Nemotron 3.5 Bolt 74B-A9B70.0%
- ~#19MiniCPM5-2BEstimated60.1%
- #19Mistral Medium 3.548.6%
No published score — estimated from related benchmarks
GDPval-AA (Elo)Verified
#60 of 91694 Elo
- #1Claude Opus 5.51866
- #2Muse Wonder Full1857
- #59Claude Haiku 4.5719
- #60MiniCPM5-2B694
- #61Nemotron 3.5 Lightning624
artificialanalysis.ai · 2026-09-24
AutomationBench-AAVerified
#63 of 841.8%
- #1Claude Fable 5.276.4%
- #2Muse Wonder Full71.3%
- #63Command A+1.8%
- #63MiniCPM5-2B1.8%
- #65Granite 4.2 30B1.5%
artificialanalysis.ai · 2026-09-24