Gemma 4 31B vs Trinity Large Thinking
Wins 5 of 6 areas
Coding · Agents · Reasoning · Long documents · Following instructions
Wins 1 of 6 areas
Facts
Gemma 4 31B is the stronger all-rounder.Trinity Large Thinking is better at facts.
Scores updated · 22 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.
- CodingWriting and fixing software40Gemma 4 31B4 of 4 tests
- AgentsCarrying out multi-step tasks on its own20Gemma 4 31B2 of 3 tests · 1 tie
- ReasoningHard problems that need careful thinking20Gemma 4 31B2 of 3 tests · 1 tie
- Long documentsFinding answers in very long texts10Gemma 4 31B1 of 1 test
- Following instructionsDoing exactly what it is asked10Gemma 4 31B1 of 1 test
- FactsGetting facts right instead of making them up01Trinity Large Thinking1 of 3 tests · 2 ties
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.
Gemma 4 31B costs 46% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Gemma 4 31B pulls ahead
- Reasons across sets of long documentsAA-LCR+31.7points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+22.8points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+19.3points ahead
Where Trinity Large Thinking pulls ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+2.5points ahead
Every test, side by side
All 22 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGemma 4 31B
- Terminal-Bench 2.1Gemma 4 31B by 22.843.420.6+22.8
- Terminal-Bench HardGemma 4 31B by 13.736.422.7+13.7
- LMArena · WebDevGemma 4 31B by 126 rating points13651239+126 rating
- SciCodeGemma 4 31B by 4.945.540.6+4.9
AgentsGemma 4 31B
- τ-Bench V3 · BankingGemma 4 31B by 914.85.8+9
- GDPValGemma 4 31B by 6.16.10+6.1
- Terminal-Bench 4.0tie00.5tie
ReasoningGemma 4 31B
- GPQA DiamondGemma 4 31B by 10.585.775.2+10.5
- Humanity's Last ExamGemma 4 31B by 7.823.615.8+7.8
- CritPttie1.40.9tie
FactsTrinity Large Thinking
- AA-Omniscience · AccuracyTrinity Large Thinking by 2.52022.5+2.5
- AA-Omniscience · Non-hallucinationtie1514.1tie
- Vectara HHEM hallucination ratelower is bettertie7.46.9tie
Following instructionsGemma 4 31B
- IFBenchGemma 4 31B by 19.375.656.3+19.3
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.
- τ²-Bench Telecom (AA run)Trinity Large Thinking by 30.259.990.1+30.2
- Artificial Analysis Coding IndexGemma 4 31B by 17.643.425.8+17.6
- vectara_avg_summary_lengthTrinity Large Thinking by 41.5 rating points75.8117.3+41.5 rating
- AA IntelligenceGemma 4 31B by 3.914.710.8+3.9
- AA-OmniscienceTrinity Large Thinking by 3.8-47.9-44.1+3.8
- vectara_answer_rateGemma 4 31B by 110099+1
- vectara_factual_consistencytie92.693.1tie
Questions people ask
Which is better, Gemma 4 31B or Trinity Large Thinking?
Gemma 4 31B wins five of the six areas where both have results: coding, agents, reasoning, long documents and following instructions. Trinity Large Thinking wins facts.
Which is better for coding?
Gemma 4 31B. It wins 4 of the 4 coding tests both models report; Trinity Large Thinking wins none.
Which is cheaper?
Gemma 4 31B costs $0.17 per million input tokens and $0.40 per million output tokens; Trinity Large Thinking costs $0.25 and $0.80. That makes Gemma 4 31B about 46% cheaper for the same work.
How do you compare the two?
We use the 22 benchmark tests both models have published scores on. The verdict counts the 15 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.