Gemma 3 4B vs LFM2.5-1.2B-Thinking
Wins 2 of 6 areas
Coding · Long documents
Wins 3 of 6 areas
Reasoning · Facts · Following instructions
LFM2.5-1.2B-Thinking wins more areas, narrowly.Gemma 3 4B is better at coding.
Scores updated · 18 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 thinking01LFM2.5-1.2B-Thinking1 of 3 tests · 2 ties
- FactsGetting facts right instead of making them up01LFM2.5-1.2B-Thinking1 of 2 tests · 1 tie
- Following instructionsDoing exactly what it is asked01LFM2.5-1.2B-Thinking1 of 1 test
- CodingWriting and fixing software10Gemma 3 4B1 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts10Gemma 3 4B1 of 1 test
- AgentsCarrying out multi-step tasks on its own00Even0 each · 1 tie
Agents, 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 LFM2.5-1.2B-Thinking pulls ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+13.5points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+4.8points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+3.8points ahead
Where Gemma 3 4B pulls ahead
- Code for real scientific research problemsSciCode+3.1points ahead
Every test, side by side
All 18 tests both models report. The winning score is in its model's colour; marks a score checked independently.
ReasoningLFM2.5-1.2B-Thinking
- GPQA DiamondLFM2.5-1.2B-Thinking by 4.829.133.9+4.8
- Humanity's Last Examtie5.36.2tie
- CritPttie00tie
FactsLFM2.5-1.2B-Thinking
- AA-Omniscience · Non-hallucinationLFM2.5-1.2B-Thinking by 3.81.85.6+3.8
- AA-Omniscience · Accuracytie7.77.4tie
Following instructionsLFM2.5-1.2B-Thinking
- IFBenchLFM2.5-1.2B-Thinking by 13.528.341.8+13.5
Other results8 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.
- MATH-500 (EM)LFM2.5-1.2B-Thinking by 14.873.288+14.8
- τ²-Bench Telecom (AA run)LFM2.5-1.2B-Thinking by 14.6519.6+14.6
- MMLU-ProLFM2.5-1.2B-Thinking by 6.143.649.6+6.1
- AA Agentic IndexLFM2.5-1.2B-Thinking by 4.81.76.5+4.8
- GSM8KGemma 3 4B by 3.689.285.6+3.6
- AA-OmniscienceLFM2.5-1.2B-Thinking by 3-82.9-80+3
- IFEvalGemma 3 4B by 1.890.288.4+1.8
- AA Intelligencetie4.85.2tie
Questions people ask
Which is better, Gemma 3 4B or LFM2.5-1.2B-Thinking?
LFM2.5-1.2B-Thinking wins three of the six areas where both have results: reasoning, facts and following instructions. Gemma 3 4B wins coding and long documents. They are level on agents.
Which is better for coding?
Gemma 3 4B. It wins 1 of the 2 coding tests both models report; LFM2.5-1.2B-Thinking wins none, and 1 is a tie.
How do you compare the two?
We use the 18 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 8 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.