Gemma 3 1B Instruct vs Gemma 3 4B
Wins 0 of 5 areas
—
Wins 4 of 5 areas
Coding · Reasoning · Long documents · Following instructions
Gemma 3 4B is the stronger all-rounder.
Scores updated · 30 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.
- ReasoningHard problems that need careful thinking01Gemma 3 4B1 of 3 tests · 2 ties
- CodingWriting and fixing software01Gemma 3 4B1 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts01Gemma 3 4B1 of 1 test
- Following instructionsDoing exactly what it is asked01Gemma 3 4B1 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 Gemma 3 4B pulls ahead
- Recent programming contest problemsLiveCodeBench v6+14.6points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+8.4points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+5.4points ahead
Where Gemma 3 1B Instruct pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+15.9points ahead
Every test, side by side
All 30 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGemma 3 4B
- LiveCodeBench v6Gemma 3 4B by 14.64.318.9+14.6
- Terminal-Bench Hardtie00.8tie
ReasoningGemma 3 4B
- GPQA DiamondGemma 3 4B by 5.423.729.1+5.4
- CritPttie00tie
- Humanity's Last Examtie5.35.3tie
FactsEven
- AA-Omniscience · Non-hallucinationGemma 3 1B Instruct by 15.917.71.8+15.9
- AA-Omniscience · AccuracyGemma 3 4B by 3.93.87.7+3.9
Following instructionsGemma 3 4B
- IFBenchGemma 3 4B by 8.419.928.3+8.4
Other results21 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.
- MGSMGemma 3 4B by 43.743.687.3+43.7
- BBHGemma 3 4B by 33.139.172.2+33.1
- HumanEvalGemma 3 4B by 29.841.571.3+29.8
- MMLU-ProGemma 3 4B by 28.914.743.6+28.9
- MATH-500 (EM)Gemma 3 4B by 2845.273.2+28
- MBPPGemma 3 4B by 2835.263.2+28
- Creative Writing v3Gemma 3 4B by 26.941.768.6+26.9
- GSM8KGemma 3 4B by 26.462.889.2+26.4
- HumanEval+Gemma 3 4B by 25.637.262.8+25.6
- Global-MMLU-LiteGemma 3 4B by 20.334.254.5+20.3
- MMLUGemma 3 4B by 18.340.158.4+18.3
- MMMLUGemma 3 4B by 15.734.450.1+15.7
- LCB v5Gemma 3 4B by 14.74.419.1+14.7
- LiveCodeBenchGemma 3 4B by 10.71.912.6+10.7
- IFEvalGemma 3 4B by 1080.290.2+10
- AA-OmniscienceGemma 3 1B Instruct by 7.5-75.5-82.9+7.5
- τ²-Bench Telecom (AA run)Gemma 3 1B Instruct by 5.510.55+5.5
- BIG-Bench Extra HardGemma 3 4B by 3.87.211+3.8
- AA Agentic IndexGemma 3 1B Instruct by 1.83.51.7+1.8
- SimpleQAGemma 3 4B by 1.82.24+1.8
- AA Intelligencetie4.84.8tie
Questions people ask
Which is better, Gemma 3 1B Instruct or Gemma 3 4B?
Gemma 3 4B wins four of the five areas where both have results: coding, reasoning, long documents and following instructions. Gemma 3 1B Instruct wins none. They are level on facts.
Which is better for coding?
Gemma 3 4B. It wins 1 of the 2 coding tests both models report; Gemma 3 1B Instruct wins none, and 1 is a tie.
How do you compare the two?
We use the 30 benchmark tests both models have published scores on. The verdict counts the 9 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 21 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.