LFM2-2.6B vs Llama 3.2 Instruct 1B
Wins 2 of 5 areas
Coding · Reasoning
Wins 2 of 5 areas
Long documents · Following instructions
The two are evenly matched.LFM2-2.6B is better at coding and reasoning; Llama 3.2 Instruct 1B at long documents and following instructions.
Scores updated · 23 tests both models report · How we compare
Where each one wins
Tests won in each of the five 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.
- CodingWriting and fixing software10LFM2-2.6B1 of 3 tests · 2 ties
- ReasoningHard problems that need careful thinking10LFM2-2.6B1 of 3 tests · 2 ties
- Long documentsFinding answers in very long texts01Llama 3.2 Instruct 1B1 of 1 test
- Following instructionsDoing exactly what it is asked01Llama 3.2 Instruct 1B1 of 1 test
- 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 tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where LFM2-2.6B pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+11points ahead
- Recent programming contest problemsLiveCodeBench v6+10.3points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+3.4points ahead
Where Llama 3.2 Instruct 1B pulls ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+3.3points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+1.6points ahead
Every test, side by side
All 23 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingLFM2-2.6B
- LiveCodeBench v6LFM2-2.6B by 10.314.44.1+10.3
- SciCodetie2.51.7tie
- Terminal-Bench Hardtie0.80tie
ReasoningLFM2-2.6B
- GPQA DiamondLFM2-2.6B by 1130.619.6+11
- CritPttie00tie
- Humanity's Last Examtie5.55.5tie
FactsEven
- AA-Omniscience · Non-hallucinationLFM2-2.6B by 3.437.133.7+3.4
- AA-Omniscience · AccuracyLlama 3.2 Instruct 1B by 1.65.47+1.6
Long documentsLlama 3.2 Instruct 1B
- AA-LCRLlama 3.2 Instruct 1B by 6.706.7+6.7
Following instructionsLlama 3.2 Instruct 1B
- IFBenchLlama 3.2 Instruct 1B by 3.319.522.8+3.3
Other results13 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.
- GSM8KLFM2-2.6B by 46.782.435.7+46.7
- MGSMLFM2-2.6B by 45.274.329.1+45.2
- MATH-500 (EM)LFM2-2.6B by 40.263.623.4+40.2
- HumanEval+LFM2-2.6B by 34.757.923.2+34.7
- IFEvalLFM2-2.6B by 27.279.652.4+27.2
- MMLULFM2-2.6B by 17.864.446.6+17.8
- MMMLULFM2-2.6B by 17.255.438.1+17.2
- τ²-Bench Telecom (AA run)LFM2-2.6B by 13.513.50+13.5
- LCB v5LFM2-2.6B by 10.814.43.6+10.8
- MMLU-ProLFM2-2.6B by 5.22620.8+5.2
- AA Agentic IndexLFM2-2.6B by 4.54.50+4.5
- AA-Omnisciencetie-54.1-54.7tie
- AA Intelligencetie5.24.8tie
Questions people ask
Which is better, LFM2-2.6B or Llama 3.2 Instruct 1B?
LFM2-2.6B and Llama 3.2 Instruct 1B each win two of the five areas where both have results. LFM2-2.6B wins coding and reasoning; Llama 3.2 Instruct 1B wins long documents and following instructions. They are level on facts.
Which is better for coding?
LFM2-2.6B. It wins 1 of the 3 coding tests both models report; Llama 3.2 Instruct 1B wins none, and 2 are ties.
How do you compare the two?
We use the 23 benchmark tests both models have published scores on. The verdict counts the 10 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 13 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.