BenchLeader

DeepSeek V4 Flash vs GLM 5.3

Full profiles: DeepSeek V4 Flash (DeepSeek) and GLM 5.3 (Zhipu AI): every benchmark, price and rank.

Updated · Data from 11 sources

As of 16 Nov 2026, DeepSeek V4 Flash and GLM 5.3 are neck and neck on quality (57.2 vs 58.8 on the BenchLeader Index, inside the ±5.0 margin), and DeepSeek V4 Flash is 8.2× cheaper at $0.262 vs $2.15 per 1M tokens blended. Choose DeepSeek V4 Flash for reasoning (59 vs 55), maths (62 vs 60), and knowledge (55 vs 54) and fast first answers (0.7 s vs 57.6 s); choose GLM 5.3 for human preference (67 vs 62) and coding (60 vs 55).

Key facts (as of 16 Nov 2026)

  • DeepSeek V4 Flash scores 57.2 on the BenchLeader Index and GLM 5.3 scores 58.8 (BenchLeader, 16 Nov 2026).
  • DeepSeek V4 Flash costs $0.262 per 1M tokens blended (list price, 16 Nov 2026).
  • GLM 5.3 costs $2.15 per 1M tokens blended (list price, 16 Nov 2026).
  • Of the 33 benchmarks both models have, GLM 5.3 leads 27 (BenchLeader, 16 Nov 2026).
  • On Arena WebDev, DeepSeek V4 Flash is rated 1582 and GLM 5.3 1623 (LMArena, as published by 16 Nov 2026).

Which to pick for…

The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.

Use casePickWhyConfidence
CodingGLM 5.360 vs 55 category score; Arena WebDev 1623 vs 1582Medium
MathsDeepSeek V4 Flash62 vs 60 category score; Arena Maths 1441 vs 1501Medium
ReasoningDeepSeek V4 Flash59 vs 55 category score; Arena Hard Prompts 1459 vs 1507Medium
KnowledgeDeepSeek V4 Flash55 vs 54 category score; MMLU-Pro 86.2% vs 86.8%Medium
Chat quality (human preference)GLM 5.367 vs 62 on Arena votesMedium
Lowest costDeepSeek V4 Flash$0.262 vs $2.15 per 1M blended; agentic coding session $0.011 vs $0.102–
Fastest first answerDeepSeek V4 Flash0.7 s (DeepSeek V4 Flash (high)) vs 57.6 s (GLM 5.3 (max)) to the first answer token–
High volume on a budgetNeither flagshipA smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs–

Key differences

  • DeepSeek V4 Flash (high) and GLM 5.3 (max) are neck and neck on quality (57.2 vs 58.8, within the ±5.0 margin).
  • DeepSeek V4 Flash (high) is stronger in knowledge, maths, reasoning.
  • GLM 5.3 (max) is stronger in coding, human preference.
  • DeepSeek V4 Flash (high) is 8.2× cheaper ($0.262 vs $2.15 per 1M blended).

At a glance

SpecDeepSeek V4 FlashGLM 5.3
DeveloperDeepSeekZhipu AI
Released31 Jul 202618 Aug 2026
Context window (tokens)1,048,5761,000,000
Compared atDeepSeek V4 Flash (high)GLM 5.3 (max)
Open weightsYesYes
Price in / out / cached input per 1M$0.150 / $0.600 / $0.0030$1.40 / $4.40 / $0.260

Category by category

Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.

CategoryDeepSeek V4 Flash (high)GLM 5.3 (max)Leader
Reasoning59.254.7DeepSeek V4 Flash
Coding55.059.7GLM 5.3
Agents & tools–57.6not comparable
Maths61.960.0DeepSeek V4 Flash
Knowledge54.853.5DeepSeek V4 Flash
Human preference62.067.2GLM 5.3
Long context–64.5not comparable

On one chart

Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.

2550 avg75ReasoningCodingMathsKnowledgeHumanpreferenceDeepSeek V4 Flash, Reasoning: 59.2DeepSeek V4 Flash, Coding: 55.0DeepSeek V4 Flash, Maths: 61.9DeepSeek V4 Flash, Knowledge: 54.8DeepSeek V4 Flash, Human preference: 62.0GLM 5.3, Reasoning: 54.7GLM 5.3, Coding: 59.7GLM 5.3, Maths: 60.0GLM 5.3, Knowledge: 53.5GLM 5.3, Human preference: 67.2

Not drawn, measured for fewer than two of these models: Agents & tools, Instruction following, Long context, Multimodal.

Model, click to add or removeIndex

Where sources disagree

Across the 33 benchmarks both have, DeepSeek V4 Flash leads 4, GLM 5.3 leads 27 and 2 are tied. Every overall index that rates both puts GLM 5.3 ahead.

Overall indexDeepSeek V4 Flash (high)GLM 5.3 (max)Leader
BenchLeader Index57.258.8neck and neck
Arena Text14391482GLM 5.3

Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).

Price, speed and latency

List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.

MeasureDeepSeek V4 Flash (high)GLM 5.3 (max)
Blended price per 1M tokens$0.262$2.15
Input / output per 1M$0.150 / $0.600$1.40 / $4.40
Cached input per 1M$0.0030$0.260
Price per task: chat reply$0.0002$0.0019
Price per task: summarize a 30-page report$0.0022$0.019
Price per task: code edit$0.0018$0.015
Price per task: agentic coding session$0.011$0.102
Price per task: structured extraction$0.0004$0.0037
Agentic coding session, input mostly cached$0.0048$0.050
Cost per benchmark task (AA Index)–$5.01
Output tokens per task (AA Index)–119k
Of which reasoning–85k
Time per task (AA Index)–23 min
Output speed82 tok/s88 tok/s
Time to first answer0.7 s57.6 s
Time to a complete answer, last measured 13 Nov 2026–144.1 s

Each model is compared at its best-ranked reasoning-effort setting. DeepSeek V4 Flash (high) scores 57.2 and takes 1 s to its first answer token; DeepSeek V4 Flash (max) scores 56.8 in 28 s, DeepSeek V4 Flash scores 53.8 in 1 s. GLM 5.3 (max) scores 58.8 and takes 58 s to its first answer token; GLM 5.3 (low) scores 47.3 in 7 s. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.

Every benchmark

149 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.

MetricPublisherDeepSeek V4 Flash (high)GLM 5.3 (max)
BenchLeader IndexBenchLeader57.258.8
Coding scoreBenchLeader55.059.7
Human preference scoreBenchLeader62.067.2
Knowledge scoreBenchLeader54.853.5
Maths scoreBenchLeader61.960.0
Reasoning scoreBenchLeader59.254.7
Agents & tools scoreBenchLeader–57.6
Long context scoreBenchLeader–64.5
Blended price $/MBenchLeader$0.262$2.15
Output speedBenchLeader82 tok/s88 tok/s
Time to first answerBenchLeader0.7 s57.6 s
Context windowBenchLeader1,048,576 tokens1,000,000 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub91.0%†90.9%
SimpleBenchSimpleBench61.1%†66.2%†
Arena Hard PromptsArena (LMArena)14591507
LiveBench 1.0 ReasoningLiveBench86.6%†85.8%†
GPQA Diamond (AA)Artificial Analysis90.8%†91.7%
Humanity's Last Exam (AA)Artificial Analysis38.5%†42.3%
HLE-Diamond (AA)Artificial Analysis–17.9%
GPQA Diamond (Vals)Vals AI89.9%88.1%
Kagi LLM BenchmarkKagi LLM Benchmark52.2%†–
ARC-AGI-1ARC Prize87.0%–
ARC-AGI-2ARC Prize56.0%–
CritPtArtificial Analysis16.6%†19.1%
CritPt v2 hard subset (AA)Artificial Analysis–33.5%
MysteryMechanismVals AI–23.0%
HLE-Diamond (Scale)Scale AI SEAL–16.4%†
Arena ExpertArena (LMArena)14671522
Chess PuzzlesEpoch AI Benchmarking Hub32.0%†37.0%
Mystery Game PuzzlesEpoch AI Benchmarking Hub34.0%†33.0%
LMCAEpoch AI Benchmarking Hub41.7%†55.5%†
DTBenchEpoch AI Benchmarking Hub90.9%†87.7%†
BTF-3Epoch AI Benchmarking Hub–14.9%†
Coding
SciCodeSciCode42.0%59.0%
WeirdMLWeirdML57.0%75.4%
FrontierCodeCognition18.8%†40.1%
Arena CodingArena (LMArena)14801522
Show all 149 benchmarks
MetricPublisherDeepSeek V4 Flash (high)GLM 5.3 (max)
Coding
Arena WebDevArena (LMArena)15821623
LiveBench 1.0 CodingLiveBench75.0%†79.0%†
SciCode (AA)Artificial Analysis50.4%†59.0%
LiveCodeBenchVals AI87.3%80.5%
IOIVals AI32.7%68.4%
SWE-bench (Vals)Vals AI88.8%95.4%
Code MigrationVals AI38.6%44.2%
Code Migration v1.1Vals AI–55.1%
ProgramBenchVals AI0.0%1.5%
Vibe Code Bench 1-100Vals AI–20.0%
Vibe Code Bench v1.1Vals AI74.7%78.1%
Vibe Code Bench 1-100 v1.1Vals AI–29.8%
SWE Atlas v1.1Scale AI SEAL–58.1%†
SWE Atlas v1.1: Codebase QnAScale AI SEAL–70.2%†
SWE Atlas v1.1: Test WritingScale AI SEAL–55.6%†
SWE Atlas v1.1: RefactoringScale AI SEAL–48.6%†
SWE-Bench Pro V2Scale AI SEAL–95.6%
SWE-Bench Pro V2 HARDScale AI SEAL–84.3%
DeepSWE v1.1Epoch AI Benchmarking Hub–69.0%
CursorBenchCursor–42.6%
ALE-BenchEpoch AI Benchmarking Hub678.21317.4†
FrontierSWEEpoch AI Benchmarking Hub–30.2%
Terminal-Bench 4.0 (AA)Artificial Analysis12.1%†41.9%
Terminal-Bench 2.1 (AA)Artificial Analysis78.7%†83.9%
Startrise FrontendStartrise–73.2%†
GameBench 2 OverallDuelLab GameBench 234.1†64.6†
GameBench 2 by reasoning settingDuelLab GameBench 243.472.3
Design Arena WebsiteDesign Arena1199†1279†
FrontierCode 1.1 (AA)Artificial Analysis–40.1%
FrontierCode 2.0 (AA)Artificial Analysis–21.6%
Agents & tools
Terminal-BenchTerminal-Bench–41.2%
APEX-AgentsMercor–56.6%†
Arena AgentArena (LMArena)–1
LiveBench Agentic CodingLiveBench46.8%†60.9%†
Terminal-Bench 2.1 (Vals)Vals AI67.0%†71.5%
MCP AtlasScale AI SEAL–84.2%†
HiL-BenchScale AI SEAL47.8%†58.3%†
GDPval-AA v2.1, normalised scoreArtificial Analysis–57.5%
GDPval-AA v2.1Artificial Analysis1437†1655
τ³-Banking (AA)Artificial Analysis39.4%†50.3%
ITBench SRE (AA)Artificial Analysis–46.1%
Analyst Agent (AA)Artificial Analysis–38.8%
CyberBenchVals AI72.1%72.1%
Finance Agent v2Vals AI49.5%55.8%
Finance Agent v2.1Vals AI–65.0%
Harvey's Legal Agent Benchmark v1.1Vals AI–14.6%
Harvey's Legal Agent Benchmark v1Vals AI8.3%8.3%
Legal Research BenchVals AI30.3%49.0%
SkillsBenchVals AI50.7%47.5%
SREBenchVals AI–9.2%
Tax Agent BenchVals AI28.1%43.3%
Terminal-Bench 4.0 (Vals)Vals AI18.7%38.9%
Terminal-Bench ScienceVals AI4.3%5.7%
Arena Agent Task CompletionArena (LMArena)–3.1
PostTrainBenchEpoch AI Benchmarking Hub–34.5%
Vending-Bench 2Andon Labs–8163.6†
AutomationBenchArtificial Analysis54.0%†62.2%
GDP.pdf (AA)Artificial Analysis11.0%†11.2%
Harvey LAB-AA v1.1Harvey–0.3%
EnterpriseOps-GymArtificial Analysis–36.4%
AA-Briefcase v1.1Artificial Analysis1255†1501
GDPxGDPx by SemiAnalysis–2.7%
Design Arena AgenticDesign Arena–1239†
Terminal-Bench-Science 0.1 (AA)Artificial Analysis–9.5%
CWE-Bench (AA)Artificial Analysis–55.8%
DeepSec-Bench (AA)Artificial Analysis–24.3%
CyberGym end-to-end (AA)Artificial Analysis–29.0%
FACTS SearchGoogle–87.7%†
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub57.5%†68.8%
FrontierMath Tier 4Epoch AI Benchmarking Hub24.4%†29.3%
OTIS Mock AIMEEpoch AI Benchmarking Hub94.4%†91.1%
ProofBenchVals AI56.0%†49.0%
LiveBench MathematicsLiveBench86.8%†87.9%†
AIME 2026MathArena95.8%†–
HMMT February 2026MathArena93.9%†–
MathArena ApexMathArena27.1%†–
Arena MathsArena (LMArena)14411501
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub33.6%†41.0%
LiveBench 1.0 Data AnalysisLiveBench79.3%†70.2%†
AA-Omniscience v1.1Artificial Analysis–6.0
MMLU-ProVals AI86.2%86.8%
LegalBenchVals AI77.7%84.8%
CorpFinVals AI61.9%–
TaxEvalVals AI70.7%72.4%
BioMysteryBenchVals AI64.4%–
Excel Modeling BenchmarkVals AI57.0%56.3%
Excel Modeling Benchmark v1.1Vals AI–69.1%
MedCodeVals AI41.4%42.9%
MedScribeVals AI80.4%88.8%
Public Benefits Bench v1.1Vals AI–68.5%
Access to Justice BenchmarkVals AI–73.7%
Financial Coaching BenchVals AI–74.1%
Immigration BenchVals AI–66.7%
AA-Omniscience v1.1: accuracyArtificial Analysis–32.3%
AA-Omniscience v1.1: non-hallucinationArtificial Analysis–61.1%
HealthBench Professional (AA)OpenAI–52.8%
FACTS ParametricGoogle–40.3%†
Instruction following
LiveBench LanguageLiveBench79.2%†79.9%†
LiveBench Instruction FollowingLiveBench65.5%†69.3%†
Arena Instruction FollowingArena (LMArena)14341481
Human preference
Arena TextArena (LMArena)14391482
EQ-Bench Creative Writing v3EQ-Bench1559†2076†
Arena Creative WritingArena (LMArena)14111456
Arena Multi-turnArena (LMArena)14471482
Arena Longer QueriesArena (LMArena)14491488
Multimodal
Code2Video-AAArtificial Analysis–1034
AA-3DArtificial Analysis–794
Long context
AA-LCR v1.1Artificial Analysis79.7%†79.7%
MLCRArtificial Analysis–17.2%
FACTS GroundingGoogle–55.2%†
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub154.4†157.1†
LiveBench 1.0LiveBench74.2%†76.1%†
AA Intelligence Index v5.0.1Artificial Analysis–31.6
AA Intelligence Index v4.3.2Artificial Analysis34.3†44.8
AA Progress Index v1Artificial Analysis309.4†322.7
Vals IndexVals AI–59.9
Vals Finance IndexVals AI–61.4%
Vals RSI IndexVals AI–22.9%
Vals Router IndexVals AI–59.7%†
Vals Legal IndexVals AI–74.6%
AA Cyber IndexArtificial Analysis–36.4
Safety & honesty
FORTRESSScale AI SEAL–28.2%
DistressBenchScale AI SEAL–72.5%†
Metadata
AA Openness IndexArtificial Analysis44.4†33.3

Data as of 2026-11-16. Best configuration of each model; every score links to its source on the model pages.

DuelLab GameBench 2, release gb2-20261111-7b598bb13de4 (https://benchmarks.duellab.org/releases/gb2-20261111-7b598bb13de4/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.

What changed

  • DeepSeek V4 Flash (high): 1 score revision.
  • GLM 5.3 (max): 1 new result (PostTrainBench).
  • GLM 5.3 (max): 4 score revisions.
  • DeepSeek V4 Flash (high): 1 score revision.
  • GLM 5.3 (max): 1 score revision; cached input price moved from $0.30/M to $0.26/M (13% down).
  • GLM 5.3 (max): cached input price moved from $0.26/M to $0.30/M (15% up).
  • DeepSeek V4 Flash (high): 1 new result (GameBench 2 by reasoning setting).
  • GLM 5.3 (max): cached input price moved from $0.30/M to $0.26/M (13% down).
  • GLM 5.3 (max): cached input price moved from $0.26/M to $0.30/M (15% up).
  • GLM 5.3 (max): BenchLeader Index 61.3 → 58.9.

Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs

DeepSeek V4 Flash vs GLM 5.3: questions

Is DeepSeek V4 Flash better than GLM 5.3?
DeepSeek V4 Flash (high) and GLM 5.3 (max) are neck and neck on quality (57.2 vs 58.8, within the ±5.0 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 16 Nov 2026, so check the category scores for your use.
DeepSeek V4 Flash vs GLM 5.3: which is cheaper, and will that change?
DeepSeek V4 Flash is cheaper today: $0.262 against $2.15 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is DeepSeek V4 Flash or GLM 5.3 better for coding?
GLM 5.3 scores higher in coding (60 vs 55 on the category index, where 50 is average). On Arena WebDev, the top shared coding benchmark, it is 1582 for DeepSeek V4 Flash against 1623 for GLM 5.3.
Which is faster, DeepSeek V4 Flash or GLM 5.3?
GLM 5.3 streams faster (88 vs 82 output tokens per second, Artificial Analysis 7-day median).
Which is better for long documents?
Both take about 1M tokens: DeepSeek V4 Flash 1,048,576, GLM 5.3 1,000,000.
More questions (3)
Which has lower latency?
DeepSeek V4 Flash answers first: 0.7 s against 57.6 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
Do independent benchmarks agree on DeepSeek V4 Flash vs GLM 5.3?
Across the 33 benchmarks both have, DeepSeek V4 Flash leads 4, GLM 5.3 leads 27 and 2 are tied. Every overall index that rates both puts GLM 5.3 ahead.
When was GLM 5.3 released?
GLM 5.3 was released on 18 Aug 2026.