Skip to content
VECTOR WIREAI INTELLIGENCE
UTC

DeepSeek-V4.1-Flash vs Qwen3.8 27B

DeepSeek · released

Wins 5 of 8 areas

Coding · Agents · Reasoning · Math · Long documents

Alibaba · released

Wins 1 of 8 areas

Following instructions

DeepSeek-V4.1-Flash is the stronger all-rounder.Qwen3.8 27B is better at following instructions.

Scores updated · 41 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
Qwen3.8 27B$0.50 to read · $3.00 to write$3.50Price from Alibaba's own price page

DeepSeek-V4.1-Flash costs 57% less for the same work.

The biggest differences

The three tests each model wins by the widest margin. Scores are out of 100.

Where DeepSeek-V4.1-Flash pulls ahead

  • Abstract visual puzzles that people can solveARC-AGI-2+35.7points aheadDeepSeek-V4.1-Flash78.1Qwen3.8 27B42.4
  • Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+30.8points aheadDeepSeek-V4.1-Flash46.4Qwen3.8 27B15.6
  • Complex command-line tasks across many fieldsTerminal-Bench 4.0+28.2points aheadDeepSeek-V4.1-Flash33.8Qwen3.8 27B5.6

Where Qwen3.8 27B pulls ahead

Every test, side by side

All 41 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
  • Terminal-Bench 4.0DeepSeek-V4.1-Flash by 28.233.85.6+28.2
  • GDPValDeepSeek-V4.1-Flash by 95546+9

Full agents ranking

ReasoningDeepSeek-V4.1-Flash

Full reasoning ranking

FactsEven

Full facts ranking

Images and chartsEven

Full images and charts ranking

MathDeepSeek-V4.1-Flash

Full math ranking

Long documentsDeepSeek-V4.1-Flash

Full long documents ranking

Following instructionsQwen3.8 27B

Full following instructions ranking

Other results19 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.

  • CritPt v2 FullDeepSeek-V4.1-Flash by 4050.610.6+40
  • JobBenchDeepSeek-V4.1-Flash by 35.168.533.4+35.1
  • DeepSWE 1.1DeepSeek-V4.1-Flash by 3274.242.2+32
  • GDPval-AA 2.1DeepSeek-V4.1-Flash by 319 rating points17461427+319 rating
  • Terminal-Bench 4.0DeepSeek-V4.1-Flash by 25.631.25.6+25.6
  • NL2RepoDeepSeek-V4.1-Flash by 21.76442.3+21.7
  • CritPt v2 HardDeepSeek-V4.1-Flash by 21254+21
  • Terminal-Bench-Science 0.1DeepSeek-V4.1-Flash by 12.912.90+12.9
  • Agents' Last ExamQwen3.8 27B by 11.131.842.9+11.1
  • HLE-DiamondDeepSeek-V4.1-Flash by 1025.315.3+10
  • SWEBench Pro v2 (Hard)DeepSeek-V4.1-Flash by 8.682.473.8+8.6
  • livebench_languageDeepSeek-V4.1-Flash by 6.881.274.3+6.8
  • ARC-AGI-1DeepSeek-V4.1-Flash by 693.587.5+6
  • AA IntelligenceDeepSeek-V4.1-Flash by 5.839.533.7+5.8
  • Chartography originalDeepSeek-V4.1-Flash by 5.717.912.2+5.7
  • AA-OmniscienceDeepSeek-V4.1-Flash by 4.7-5.3-10+4.7
  • BabyVisionDeepSeek-V4.1-Flash by 489.685.6+4
  • livebench_data_analysisDeepSeek-V4.1-Flash by 2.779.376.6+2.7
  • GDP (Surge AI)tie17.216.4tie

Questions people ask

Which is better, DeepSeek-V4.1-Flash or Qwen3.8 27B?

DeepSeek-V4.1-Flash wins five of the eight areas we test: coding, agents, reasoning, math and long documents. Qwen3.8 27B wins following instructions. They are level on facts and images and charts.

Which is better for coding?

DeepSeek-V4.1-Flash. It wins 5 of the 5 coding tests both models report; Qwen3.8 27B wins none.

Which is cheaper?

DeepSeek-V4.1-Flash costs $0.30 per million input tokens and $1.20 per million output tokens; Qwen3.8 27B costs $0.50 and $3.00. That makes DeepSeek-V4.1-Flash about 57% cheaper for the same work.

How do you compare the two?

We use the 41 benchmark tests both models have published scores on. The verdict counts the 22 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 19 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.