DeepSeek-V3 vs Nova Lite
Wins 4 of 6 areas
Coding · Reasoning · Long documents · Following instructions
Wins 0 of 6 areas
—
DeepSeek-V3 is the stronger all-rounder.Nova Lite is cheaper.
Scores updated · 29 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 software20DeepSeek-V32 of 2 tests
- ReasoningHard problems that need careful thinking10DeepSeek-V31 of 3 tests · 2 ties
- Long documentsFinding answers in very long texts10DeepSeek-V31 of 1 test
- Following instructionsDoing exactly what it is asked10DeepSeek-V31 of 1 test
- AgentsCarrying out multi-step tasks on its own00Even0 each · 1 tie
- FactsGetting facts right instead of making them up11Even1 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.
Nova Lite costs 74% 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 DeepSeek-V3 pulls ahead
- Code for real scientific research problemsSciCode+25.1points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+22.2points ahead
- Reasons across sets of long documentsAA-LCR+21.7points ahead
Where Nova Lite pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+28.3points ahead
Every test, side by side
All 29 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingDeepSeek-V3
- SciCodeDeepSeek-V3 by 25.13913.9+25.1
- Terminal-Bench HardDeepSeek-V3 by 14.415.20.8+14.4
ReasoningDeepSeek-V3
- GPQA DiamondDeepSeek-V3 by 22.265.543.3+22.2
- Humanity's Last Examtie4.74.3tie
- CritPttie00tie
FactsEven
- AA-Omniscience · Non-hallucinationNova Lite by 28.314.142.4+28.3
- AA-Omniscience · AccuracyDeepSeek-V3 by 15.725.49.8+15.7
- Vectara HHEM hallucination ratelower is bettertie6.16.1tie
Following instructionsDeepSeek-V3
- IFBenchDeepSeek-V3 by 6.94134.1+6.9
Other results18 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)DeepSeek-V3 by 29.647.117.5+29.6
- HumanEvalNova Lite by 20.265.285.4+20.2
- MMLUDeepSeek-V3 by 19.288.569.3+19.2
- Artificial Analysis Coding IndexDeepSeek-V3 by 17.9235.1+17.9
- MATHDeepSeek-V3 by 12.390.277.9+12.3
- naturalquestions_closedbookDeepSeek-V3 by 11.546.735.2+11.5
- DROPDeepSeek-V3 by 11.491.680.2+11.4
- vectara_avg_summary_lengthNova Lite by 10.181.791.8+10.1
- BBHDeepSeek-V3 by 5.187.582.4+5.1
- IFEvalNova Lite by 3.686.189.7+3.6
- AA IntelligenceDeepSeek-V3 by 39.76.7+3
- ARC-ChallengeDeepSeek-V3 by 2.995.392.4+2.9
- NarrativeQADeepSeek-V3 by 2.879.676.8+2.8
- OpenBookQADeepSeek-V3 by 2.695.492.8+2.6
- vectara_answer_rateNova Lite by 2.497.599.9+2.4
- AA-OmniscienceDeepSeek-V3 by 1.5-40.7-42.2+1.5
- GSM8Ktie9494.5tie
- vectara_factual_consistencytie93.993.9tie
Questions people ask
Which is better, DeepSeek-V3 or Nova Lite?
DeepSeek-V3 wins four of the six areas where both have results: coding, reasoning, long documents and following instructions. Nova Lite wins none, but costs 74% less. They are level on agents and facts.
Which is better for coding?
DeepSeek-V3. It wins 2 of the 2 coding tests both models report; Nova Lite wins none.
Which is cheaper?
DeepSeek-V3 costs $0.24 per million input tokens and $0.90 per million output tokens; Nova Lite costs $0.06 and $0.24. That makes Nova Lite about 74% cheaper for the same work.
How do you compare the two?
We use the 29 benchmark tests both models have published scores on. The verdict counts the 11 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 18 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.