MiniCPM5-2B vs Nemotron 3.5 Lightning
Wins 1 of 6 areas
Agents
Wins 4 of 6 areas
Coding · Reasoning · Long documents · Following instructions
Nemotron 3.5 Lightning is the stronger all-rounder.MiniCPM5-2B is better at agents.
Scores updated · 20 tests both models report · How we compare
Where each one wins
Tests won in each of the six areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- ReasoningHard problems that need careful thinking02Nemotron 3.5 Lightning2 of 3 tests · 1 tie
- CodingWriting and fixing software02Nemotron 3.5 Lightning2 of 2 tests
- Long documentsFinding answers in very long texts01Nemotron 3.5 Lightning1 of 1 test
- Following instructionsDoing exactly what it is asked01Nemotron 3.5 Lightning1 of 1 test
- AgentsCarrying out multi-step tasks on its own21MiniCPM5-2B2 of 4 tests · 1 tie
- FactsGetting facts right instead of making them up11Even1 each
Long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Nemotron 3.5 Lightning pulls ahead
- Multi-step tasks using many real software toolsMCP Atlas+20points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+15.7points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+9.6points ahead
Where MiniCPM5-2B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+15.7points ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+11.9points ahead
- Real work tasks from 44 professionsGDPVal+3.5points ahead
Every test, side by side
All 20 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingNemotron 3.5 Lightning
- Terminal-Bench 2.1Nemotron 3.5 Lightning by 15.78.624.3+15.7
- SciCodeNemotron 3.5 Lightning by 5.826.332.1+5.8
AgentsMiniCPM5-2B
- MCP AtlasNemotron 3.5 Lightning by 2026.446.4+20
- τ-Bench V3 · BankingMiniCPM5-2B by 11.920.88.9+11.9
- GDPValMiniCPM5-2B by 3.510.46.9+3.5
- Terminal-Bench 4.0tie00.5tie
ReasoningNemotron 3.5 Lightning
- GPQA DiamondNemotron 3.5 Lightning by 4.170.274.3+4.1
- Humanity's Last ExamNemotron 3.5 Lightning by 1.78.910.6+1.7
- CritPttie0.30tie
FactsEven
- AA-Omniscience · Non-hallucinationMiniCPM5-2B by 15.778.162.4+15.7
- AA-Omniscience · AccuracyNemotron 3.5 Lightning by 68.414.4+6
Long documentsNemotron 3.5 Lightning
- AA-LCRNemotron 3.5 Lightning by 159.360.3+1
Following instructionsNemotron 3.5 Lightning
- IFBenchNemotron 3.5 Lightning by 9.662.171.7+9.6
Other results7 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- WebArena-VerifiedNemotron 3.5 Lightning by 19.69.128.7+19.6
- Artificial Analysis Coding IndexNemotron 3.5 Lightning by 12.314.526.8+12.3
- BFCLv4MiniCPM5-2B by 8.962.653.8+8.9
- IFEvalNemotron 3.5 Lightning by 6.487.894.2+6.4
- AA-OmniscienceMiniCPM5-2B by 6.1-11.7-17.7+6.1
- AA Agentic IndexMiniCPM5-2B by 2.996.1+2.9
- AA Intelligencetie12.512.9tie
Questions people ask
Which is better, MiniCPM5-2B or Nemotron 3.5 Lightning?
Nemotron 3.5 Lightning wins four of the six areas where both have results: coding, reasoning, long documents and following instructions. MiniCPM5-2B wins agents. They are level on facts.
Which is better for coding?
Nemotron 3.5 Lightning. It wins 2 of the 2 coding tests both models report; MiniCPM5-2B wins none.
How do you compare the two?
We use the 20 benchmark tests both models have published scores on. The verdict counts the 13 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 7 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.