Skip to content
VECTOR WIREAI INTELLIGENCE
UTC

DeepSeek-V4.1-Flash vs Gemma 4 31B

DeepSeek · released

Wins 6 of 6 areas

Coding · Agents · Reasoning · Facts · Images and charts · Long documents

Google · released

Wins 0 of 6 areas

—

DeepSeek-V4.1-Flash is the stronger all-rounder.Gemma 4 31B is cheaper.

Scores updated · 33 tests both models report · How we compare

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.

DeepSeek-V4.1-Flash$0.30 to read · $1.20 to write$1.50Price from Artificial Analysis
Gemma 4 31B$0.15 to read · $0.40 to write$0.55Price from Artificial Analysis

Gemma 4 31B costs 63% 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-V4.1-Flash pulls ahead

  • Real work tasks from 44 professionsGDPVal+49points aheadDeepSeek-V4.1-Flash55Gemma 4 31B6
  • Command-line tasks in a real terminalTerminal-Bench 2.1+47.2points aheadDeepSeek-V4.1-Flash90.6Gemma 4 31B43.4
  • Short factual questions, answered correctlySimpleQA Verified+32.5points aheadDeepSeek-V4.1-Flash42.9Gemma 4 31B10.4

Where Gemma 4 31B pulls ahead

Every test, side by side

All 33 tests both models report. The winning score is in its model's colour; marks a score checked independently.

CodingDeepSeek-V4.1-Flash

Full coding ranking

AgentsDeepSeek-V4.1-Flash
  • GDPValDeepSeek-V4.1-Flash by 49556+49
  • Terminal-Bench 4.0DeepSeek-V4.1-Flash by 33.833.80+33.8
  • AA IT-Bench SREDeepSeek-V4.1-Flash by 9.646.937.3+9.6

Full agents ranking

ReasoningDeepSeek-V4.1-Flash

Full reasoning ranking

FactsDeepSeek-V4.1-Flash

Full facts ranking

Images and chartsDeepSeek-V4.1-Flash

Full images and charts ranking

Long documentsDeepSeek-V4.1-Flash
  • LongBench v2DeepSeek-V4.1-Flash by 16.16952.9+16.1
  • AA-LCRDeepSeek-V4.1-Flash by 1382.769.7+13

Full long documents ranking

Other results16 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.

  • GDPval-AA 2.1DeepSeek-V4.1-Flash by 1119 rating points1746627+1119 rating
  • SWEBench Pro v2 (Hard)DeepSeek-V4.1-Flash by 68.282.414.2+68.2
  • JobBenchDeepSeek-V4.1-Flash by 60.368.58.2+60.3
  • NL2RepoDeepSeek-V4.1-Flash by 48.56415.5+48.5
  • AA-OmniscienceDeepSeek-V4.1-Flash by 42.6-5.3-47.9+42.6
  • HLE (with tools)DeepSeek-V4.1-Flash by 37.463.926.5+37.4
  • Terminal-Bench 4.0DeepSeek-V4.1-Flash by 31.231.20+31.2
  • AA IntelligenceDeepSeek-V4.1-Flash by 24.839.514.7+24.8
  • CritPt v2 HardDeepSeek-V4.1-Flash by 21254+21
  • HLE-DiamondDeepSeek-V4.1-Flash by 18.125.37.2+18.1
  • Terminal-Bench-Science 0.1DeepSeek-V4.1-Flash by 12.912.90+12.9
  • GDP (Surge AI)DeepSeek-V4.1-Flash by 11.217.26+11.2
  • Chartography originalDeepSeek-V4.1-Flash by 5.917.912+5.9
  • vectara_avg_summary_lengthDeepSeek-V4.1-Flash by 32.4 rating points108.275.8+32.4 rating
  • vectara_factual_consistencyDeepSeek-V4.1-Flash by 3.295.892.6+3.2
  • vectara_answer_ratetie99.9100tie

Questions people ask

Which is better, DeepSeek-V4.1-Flash or Gemma 4 31B?

DeepSeek-V4.1-Flash wins all six areas where both have results: coding, agents, reasoning, facts, images and charts and long documents. Gemma 4 31B wins none, but costs 63% less.

Which is better for coding?

DeepSeek-V4.1-Flash. It wins 3 of the 3 coding tests both models report; Gemma 4 31B wins none.

Which is cheaper?

DeepSeek-V4.1-Flash costs $0.30 per million input tokens and $1.20 per million output tokens; Gemma 4 31B costs $0.15 and $0.40. That makes Gemma 4 31B about 63% cheaper for the same work.

How do you compare the two?

We use the 33 benchmark tests both models have published scores on. The verdict counts the 17 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 16 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.