DeepSeek-V4-Pro vs Gemma 4 26B A4B
Wins 6 of 7 areas
Coding · Agents · Reasoning · Math · Long documents · Following instructions
Wins 1 of 7 areas
Facts
DeepSeek-V4-Pro is the stronger all-rounder.Gemma 4 26B A4B is cheaper and better at facts.
Scores updated · 42 tests both models report · How we compare
Where each one wins
Tests won in each of the seven 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 software80DeepSeek-V4-Pro8 of 8 tests
- AgentsCarrying out multi-step tasks on its own50DeepSeek-V4-Pro5 of 5 tests
- ReasoningHard problems that need careful thinking30DeepSeek-V4-Pro3 of 3 tests
- MathCompetition and research-level math30DeepSeek-V4-Pro3 of 3 tests
- Long documentsFinding answers in very long texts10DeepSeek-V4-Pro1 of 1 test
- Following instructionsDoing exactly what it is asked10DeepSeek-V4-Pro1 of 1 test
- FactsGetting facts right instead of making them up12Gemma 4 26B A4B2 of 3 tests
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 26B A4B costs 59% 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-V4-Pro pulls ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+63.2points ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+58.9points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+57.1points ahead
Where Gemma 4 26B A4B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+7.7points ahead
Every test, side by side
All 42 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingDeepSeek-V4-Pro
- SWE-bench VerifiedDeepSeek-V4-Pro by 63.280.617.4+63.2
- SWE-bench MultilingualDeepSeek-V4-Pro by 58.976.217.3+58.9
- SWE-bench ProDeepSeek-V4-Pro by 41.655.413.8+41.6
- Terminal-Bench HardDeepSeek-V4-Pro by 32.646.213.6+32.6
- Terminal-Bench 2.1DeepSeek-V4-Pro by 256439+25
- LiveCodeBench v6DeepSeek-V4-Pro by 15.492.577.1+15.4
- SciCodeDeepSeek-V4-Pro by 10.850.840+10.8
- LMArena · WebDevDeepSeek-V4-Pro by 105 rating points14641359+105 rating
AgentsDeepSeek-V4-Pro
- BrowseCompDeepSeek-V4-Pro by 57.183.426.3+57.1
- GDPValDeepSeek-V4-Pro by 29.532.93.4+29.5
- MCP AtlasDeepSeek-V4-Pro by 23.673.650+23.6
- τ-Bench V3 · BankingDeepSeek-V4-Pro by 18.130.112+18.1
- AA IT-Bench SREDeepSeek-V4-Pro by 14.738.323.6+14.7
ReasoningDeepSeek-V4-Pro
- Humanity's Last ExamDeepSeek-V4-Pro by 18.237.519.3+18.2
- CritPtDeepSeek-V4-Pro by 12.912.90+12.9
- GPQA DiamondDeepSeek-V4-Pro by 9.688.879.2+9.6
FactsGemma 4 26B A4B
- AA-Omniscience · AccuracyDeepSeek-V4-Pro by 23.94319.1+23.9
- AA-Omniscience · Non-hallucinationGemma 4 26B A4B by 7.75.913.6+7.7
- Vectara HHEM hallucination ratelower is betterGemma 4 26B A4B by 3.48.65.2+3.4
MathDeepSeek-V4-Pro
- HMMT Feb. 2026DeepSeek-V4-Pro by 16.295.279+16.2
- IMOAnswerBenchDeepSeek-V4-Pro by 15.589.874.3+15.5
- AIME 2026DeepSeek-V4-Pro by 8.496.788.3+8.4
Following instructionsDeepSeek-V4-Pro
- IFBenchDeepSeek-V4-Pro by 4.176.572.4+4.1
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-V4-Pro by 52.696.243.6+52.6
- GDPval-AA v2DeepSeek-V4-Pro by 499 rating points1306807+499 rating
- HLE (with tools)DeepSeek-V4-Pro by 42.86017.2+42.8
- Tool-DecathlonDeepSeek-V4-Pro by 40.852.812+40.8
- AA-OmniscienceDeepSeek-V4-Pro by 40.1-10.7-50.8+40.1
- Terminal-Bench 2.0DeepSeek-V4-Pro by 33.767.934.2+33.7
- NL2RepoDeepSeek-V4-Pro by 26.938.511.6+26.9
- Artificial Analysis Coding IndexDeepSeek-V4-Pro by 20.159.439.3+20.1
- AA Agentic IndexDeepSeek-V4-Pro by 16.727.711+16.7
- PinchBenchDeepSeek-V4-Pro by 13.988.674.7+13.9
- AA IntelligenceDeepSeek-V4-Pro by 13.730.416.7+13.7
- τ³-Bench BankingDeepSeek-V4-Pro by 122614+12
- vectara_avg_summary_lengthDeepSeek-V4-Pro by 86.7 rating points153.867.1+86.7 rating
- HMMT Nov. 2025DeepSeek-V4-Pro by 6.994.487.5+6.9
- GPQA (unspecified)DeepSeek-V4-Pro by 5.587.882.3+5.5
- MMLU-ProDeepSeek-V4-Pro by 4.987.582.6+4.9
- vectara_factual_consistencyGemma 4 26B A4B by 3.491.494.8+3.4
- vectara_answer_rateGemma 4 26B A4B by 2.697.299.8+2.6
Questions people ask
Which is better, DeepSeek-V4-Pro or Gemma 4 26B A4B?
DeepSeek-V4-Pro wins six of the seven areas where both have results: coding, agents, reasoning, math, long documents and following instructions. Gemma 4 26B A4B wins facts, and costs 59% less.
Which is better for coding?
DeepSeek-V4-Pro. It wins 8 of the 8 coding tests both models report; Gemma 4 26B A4B wins none.
Which is cheaper?
DeepSeek-V4-Pro costs $0.43 per million input tokens and $0.87 per million output tokens; Gemma 4 26B A4B costs $0.13 and $0.40. That makes Gemma 4 26B A4B about 59% cheaper for the same work.
How do you compare the two?
We use the 42 benchmark tests both models have published scores on. The verdict counts the 24 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.