DeepSeek-V4-Flash vs Kimi K3
Wins 0 of 7 areas
—
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Math · Long documents · Following instructions
Kimi K3 is the stronger all-rounder.DeepSeek-V4-Flash is cheaper.
Scores updated · 48 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.
- AgentsCarrying out multi-step tasks on its own05Kimi K35 of 6 tests · 1 tie
- CodingWriting and fixing software05Kimi K35 of 5 tests
- ReasoningHard problems that need careful thinking15Kimi K35 of 6 tests
- FactsGetting facts right instead of making them up03Kimi K33 of 3 tests
- MathCompetition and research-level math13Kimi K33 of 5 tests · 1 tie
- Long documentsFinding answers in very long texts01Kimi K31 of 1 test
- Following instructionsDoing exactly what it is asked01Kimi K31 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.
DeepSeek-V4-Flash costs 90% 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 Kimi K3 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+38.5points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+18points ahead
- Short factual questions, answered correctlySimpleQA Verified+16.5points ahead
Where DeepSeek-V4-Flash pulls ahead
- Fresh competition math problemsLiveBench · Mathematics+2.4points ahead
- Abstract visual puzzles that people can solveARC-AGI-2+1points ahead
Every test, side by side
All 48 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingKimi K3
- LMArena · WebDevKimi K3 by 228 rating points14301658+228 rating
- LiveBench · Agentic CodingKimi K3 by 15.446.862.2+15.4
- SciCodeKimi K3 by 9.250.359.5+9.2
- LiveBench · CodingKimi K3 by 6.57581.5+6.5
- Terminal-Bench 2.1Kimi K3 by 6.378.785+6.3
AgentsKimi K3
- BrowseCompKimi K3 by 1873.291.2+18
- AA IT-Bench SREKimi K3 by 16.231.547.7+16.2
- MCP AtlasKimi K3 by 15.26984.2+15.2
- τ-Bench V3 · BankingKimi K3 by 6.639.446+6.6
- GDPValKimi K3 by 4.946.951.8+4.9
- Terminal-Bench 4.0tie12.112.6tie
ReasoningKimi K3
- SimpleBenchKimi K3 by 14.446.360.7+14.4
- Humanity's Last ExamKimi K3 by 8.338.646.9+8.3
- CritPtKimi K3 by 6.816.623.4+6.8
- LiveBench · ReasoningKimi K3 by 4.186.690.7+4.1
- GPQA DiamondKimi K3 by 2.790.893.5+2.7
- ARC-AGI-2DeepSeek-V4-Flash by 161.460.4+1
FactsKimi K3
- AA-Omniscience · Non-hallucinationKimi K3 by 38.58.346.8+38.5
- SimpleQA VerifiedKimi K3 by 16.534.150.6+16.5
- AA-Omniscience · AccuracyKimi K3 by 7.240.447.6+7.2
MathKimi K3
- FrontierMath Tiers 1-3 (v2)Kimi K3 by 14.757.572.2+14.7
- FrontierMath Tier 4Kimi K3 by 14.624.439+14.6
- LiveBench · MathematicsDeepSeek-V4-Flash by 2.486.884.4+2.4
- HMMT Feb. 2026Kimi K3 by 2.294.897+2.2
- AIME 2026tie95.896.7tie
Following instructionsKimi K3
- LiveBench · Instruction FollowingKimi K3 by 5.965.571.4+5.9
Other results21 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 v2Kimi K3 by 493 rating points11891682+493 rating
- AA-OmniscienceKimi K3 by 34-14.319.7+34
- HLE (with tools)Kimi K3 by 14.745.159.8+14.7
- DeepSWE 1.1Kimi K3 by 14.654.469+14.6
- DeepSWEKimi K3 by 13.154.467.5+13.1
- JobBenchKimi K3 by 11.641.352.9+11.6
- HLE (wo / w tools)Kimi K3 by 10.945.156+10.9
- AA IntelligenceKimi K3 by 9.334.343.6+9.3
- AA Agentic IndexKimi K3 by 8.941.750.6+8.9
- Artificial Analysis Coding IndexKimi K3 by 7.169.176.2+7.1
- livebench_languageKimi K3 by 6.379.285.5+6.3
- Toolathlon VerifiedKimi K3 by 6.270.376.5+6.2
- AutomationBench PublicKimi K3 by 5.725.130.8+5.7
- ARC-AGI-1Kimi K3 by 5.58994.5+5.5
- DSBench-FullStackKimi K3 by 568.773.7+5
- NL2RepoKimi K3 by 3.854.258+3.8
- DSBench-HardKimi K3 by 3.459.663+3.4
- CyberGymKimi K3 by 3.376.780+3.3
- Agents' Last ExamKimi K3 by 3.125.228.3+3.1
- ToolathlonKimi K3 by 2.970.373.2+2.9
- livebench_data_analysistie79.378.7tie
Questions people ask
Which is better, DeepSeek-V4-Flash or Kimi K3?
Kimi K3 wins all seven areas where both have results: coding, agents, reasoning, facts, math, long documents and following instructions. DeepSeek-V4-Flash wins none, but costs 90% less.
Which is better for coding?
Kimi K3. It wins 5 of the 5 coding tests both models report; DeepSeek-V4-Flash wins none.
Which is cheaper?
DeepSeek-V4-Flash costs $0.44 per million input tokens and $1.32 per million output tokens; Kimi K3 costs $3.00 and $15.00. That makes DeepSeek-V4-Flash about 90% cheaper for the same work.
How do you compare the two?
We use the 48 benchmark tests both models have published scores on. The verdict counts the 27 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 21 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.