Gemini 2.0 Flash (Non-Reasoning) vs Llama 3 Instruct 70B
Wins 5 of 5 areas
Coding · Reasoning · Facts · Long documents · Following instructions
Wins 0 of 5 areas
—
Gemini 2.0 Flash (Non-Reasoning) is the stronger all-rounder.
Scores updated · 20 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 software20Gemini 2.0 Flash (Non-Reasoning)2 of 2 tests
- FactsGetting facts right instead of making them up20Gemini 2.0 Flash (Non-Reasoning)2 of 2 tests
- ReasoningHard problems that need careful thinking10Gemini 2.0 Flash (Non-Reasoning)1 of 3 tests · 2 ties
- Long documentsFinding answers in very long texts10Gemini 2.0 Flash (Non-Reasoning)1 of 1 test
- Following instructionsDoing exactly what it is asked10Gemini 2.0 Flash (Non-Reasoning)1 of 1 test
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 Gemini 2.0 Flash (Non-Reasoning) pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+24.4points ahead
- Code for real scientific research problemsSciCode+14.4points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+5.8points ahead
Where Llama 3 Instruct 70B pulls ahead
No clear win on a test scored out of 100.
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.
CodingGemini 2.0 Flash (Non-Reasoning)
- SciCodeGemini 2.0 Flash (Non-Reasoning) by 14.433.318.9+14.4
- Terminal-Bench HardGemini 2.0 Flash (Non-Reasoning) by 33.80.8+3
ReasoningGemini 2.0 Flash (Non-Reasoning)
- GPQA DiamondGemini 2.0 Flash (Non-Reasoning) by 24.462.337.9+24.4
- Humanity's Last Examtie4.34.5tie
- CritPttie00tie
FactsGemini 2.0 Flash (Non-Reasoning)
- AA-Omniscience · AccuracyGemini 2.0 Flash (Non-Reasoning) by 5.823.517.7+5.8
- AA-Omniscience · Non-hallucinationGemini 2.0 Flash (Non-Reasoning) by 1.313.812.5+1.3
Long documentsGemini 2.0 Flash (Non-Reasoning)
- AA-LCRGemini 2.0 Flash (Non-Reasoning) by 31.331.30+31.3
Following instructionsGemini 2.0 Flash (Non-Reasoning)
- IFBenchGemini 2.0 Flash (Non-Reasoning) by 3.140.237.1+3.1
Other results11 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)Gemini 2.0 Flash (Non-Reasoning) by 29.529.50+29.5
- AA Agentic IndexGemini 2.0 Flash (Non-Reasoning) by 15.815.80+15.8
- AA-OmniscienceGemini 2.0 Flash (Non-Reasoning) by 11.8-42.5-54.3+11.8
- HarmBenchGemini 2.0 Flash (Non-Reasoning) by 8.272.264+8.2
- Artificial Analysis Coding IndexGemini 2.0 Flash (Non-Reasoning) by 6.813.66.8+6.8
- XSTestLlama 3 Instruct 70B by 4.192.796.8+4.1
- AA IntelligenceGemini 2.0 Flash (Non-Reasoning) by 3.48.95.5+3.4
- AIR-Bench 2024Gemini 2.0 Flash (Non-Reasoning) by 2.967.564.6+2.9
- anthropic_red_teamGemini 2.0 Flash (Non-Reasoning) by 2.799.396.7+2.7
- simple_safety_testsLlama 3 Instruct 70B by 1.397.799+1.3
- bbqGemini 2.0 Flash (Non-Reasoning) by 19291+1
Questions people ask
Which is better, Gemini 2.0 Flash (Non-Reasoning) or Llama 3 Instruct 70B?
Gemini 2.0 Flash (Non-Reasoning) wins all five areas where both have results: coding, reasoning, facts, long documents and following instructions. Llama 3 Instruct 70B wins none.
Which is better for coding?
Gemini 2.0 Flash (Non-Reasoning). It wins 2 of the 2 coding tests both models report; Llama 3 Instruct 70B wins none.
How do you compare the two?
We use the 20 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 11 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.