LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

DeepSeek V3.2 vs Hy3

Updated October 21, 2026. 42 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace DeepSeek V3.2 Replace Hy3
DeepSeek V3.2DeepSeek
Coding
54.4
Overall
21.5
Output $/M
$0.42
1
Hy3Tencent
Coding
43.9
Overall
No data
Output $/M
$0.528
41
vs boards won of 42 shared

Decision reading

Pick DeepSeek V3.2 for coding: it scores 54.4 on the Vibe Coding Index against 43.9 for Hy3.

Pick Hy3 when the context window is larger, 262K against 164K tokens, or when what matters is LMArena WebDev or LMArena Text: coding, where it leads.

Model ADeepSeek V3.2DeepSeek54.4/1002 independent benchmarks$0.99 per solved taskModel BHy3Tencent43.9/1001 independent benchmarkNo cost per solved task
Shared results
42
DeepSeek V3.2 only
33
Hy3 only
2
Like-for-like categories
6 of 9

Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.

Which one for your work

Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.

At a glance

Coding index
DeepSeek V3.2 54.4 from 2 benchmarks, Hy3 43.9 from 1 benchmark.
Widest category
Coding, 14.2 points apart on average over 5 shared benchmarks.
Widest single gap
LMArena WebDev: React: DeepSeek V3.2 1350 against Hy3 1511, 26.9 points apart on a 0 to 100 scale.
Output price
DeepSeek V3.2 $0.42 against Hy3 $0.528 per million output tokens, 1.3x apart.
Context window
DeepSeek V3.2 takes 164K tokens, Hy3 262K.

Benchmark by benchmark

First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.

Boards won, where both have a result
CategoryDeepSeek V3.2BarsHy3Ties
Coding05
Reasoning05
Math02
Knowledge13
Web design08
General018
All shared boards141
CodingLike-for-likeFour of these boards feed the Vibe Coding Index05
Coding benchmarks for DeepSeek V3.2 and Hy3
BenchmarkDeepSeek V3.2Hy3Leader
SWE-bench Verified70.0%Not measuredOnly DeepSeek V3.2 measured
SWE-bench Multilingual59.0%Not measuredOnly DeepSeek V3.2 measured
LMArena WebDev13621507Hy3
AA Coding Index44.2Not measuredOnly DeepSeek V3.2 measured
LiveCodeBench (Vals AI)80.7%Not measuredOnly DeepSeek V3.2 measured
SWE-bench Verified (Vals AI)67.6%Not measuredOnly DeepSeek V3.2 measured
Vibe Code Bench (Vals AI)5.1%Not measuredOnly DeepSeek V3.2 measured
LiveCodeBench (Artificial Analysis run)86.2%Not measuredOnly DeepSeek V3.2 measured
SciCodeNot measured48.6%Only Hy3 measured
Terminal-Bench Hard (Artificial Analysis run)35.6%Not measuredOnly DeepSeek V3.2 measured
Terminal-Bench39.6%Not measuredOnly DeepSeek V3.2 measured
ALE-BenchNot measured838.6Only Hy3 measured
LMArena Text: coding14531466Hy3
LMArena Text: software and IT14491463Hy3
LMArena WebDev: React13501511Hy3
LMArena WebDev: HTML13711487Hy3
Neither model has a result on the other 22 boards here
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
Intelligence benchmarks for DeepSeek V3.2 and Hy3
BenchmarkDeepSeek V3.2Hy3Leader
AA Intelligence Index21.5Not measuredOnly DeepSeek V3.2 measured
Epoch Capabilities Index146.1Not measuredOnly DeepSeek V3.2 measured
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index3 boards, none shared
Agentic work benchmarks for DeepSeek V3.2 and Hy3
BenchmarkDeepSeek V3.2Hy3Leader
τ²-Bench Telecom90.6%Not measuredOnly DeepSeek V3.2 measured
APEX-Agents21.3%Not measuredOnly DeepSeek V3.2 measured
Vending-Bench 2$1,034Not measuredOnly DeepSeek V3.2 measured
Neither model has a result on the other 36 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index05
Reasoning benchmarks for DeepSeek V3.2 and Hy3
BenchmarkDeepSeek V3.2Hy3Leader
GPQA Diamond80.3%Not measuredOnly DeepSeek V3.2 measured
GPQA Diamond (Artificial Analysis run)84.0%89.7%Hy3
Humanity's Last Exam (Artificial Analysis run)24.6%33.5%Hy3
ARC-AGI57.0%Not measuredOnly DeepSeek V3.2 measured
ARC-AGI-24.0%Not measuredOnly DeepSeek V3.2 measured
LMArena Text: hard prompts14341447Hy3
LMArena Text: expert14361461Hy3
LMArena Text: hard prompts in English14461455Hy3
Neither model has a result on the other 13 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index02
Math benchmarks for DeepSeek V3.2 and Hy3
BenchmarkDeepSeek V3.2Hy3Leader
AIME84.6%Not measuredOnly DeepSeek V3.2 measured
MGSM (Vals AI)86.0%Not measuredOnly DeepSeek V3.2 measured
AA Math Index92.0Not measuredOnly DeepSeek V3.2 measured
AIME 2025 (Artificial Analysis run)92.0%Not measuredOnly DeepSeek V3.2 measured
LMArena Text: math14351473Hy3
LMArena Text: mathematical work14401482Hy3
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index13
Knowledge benchmarks for DeepSeek V3.2 and Hy3
BenchmarkDeepSeek V3.2Hy3Leader
MMLU Pro84.9%Not measuredOnly DeepSeek V3.2 measured
LegalBench (Vals AI)76.1%Not measuredOnly DeepSeek V3.2 measured
CaseLaw (Vals AI)55.4%Not measuredOnly DeepSeek V3.2 measured
CorpFin (Vals AI)51.0%Not measuredOnly DeepSeek V3.2 measured
TaxEval (Vals AI)68.2%Not measuredOnly DeepSeek V3.2 measured
MedQA (Vals AI)93.9%Not measuredOnly DeepSeek V3.2 measured
MMLU Pro (Artificial Analysis run)86.2%Not measuredOnly DeepSeek V3.2 measured
LMArena Text: science14461457Hy3
LMArena Text: medicine and healthcare14401452Hy3
LMArena Text: legal and government14251422DeepSeek V3.2
LMArena Text: business and finance14101424Hy3
Neither model has a result on the other 6 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index08
Web design benchmarks for DeepSeek V3.2 and Hy3
BenchmarkDeepSeek V3.2Hy3Leader
Design Arena: websites11751230Hy3
Design Arena: UI components11491205Hy3
Design Arena: game dev11321208Hy3
Design Arena: data viz11581163Hy3
Design Arena: 3D11331233Hy3
Design Arena: all code categories11641227Hy3
Design Arena: SVG10321199Hy3
Design Arena: ASCII art10821157Hy3
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index018
General benchmarks for DeepSeek V3.2 and Hy3
BenchmarkDeepSeek V3.2Hy3Leader
IFBench60.7%Not measuredOnly DeepSeek V3.2 measured
Long context reasoning (AA-LCR)73.3%79.0%Hy3
DTBench85.6%Not measuredOnly DeepSeek V3.2 measured
LMCA28.8%Not measuredOnly DeepSeek V3.2 measured
CL-bench12.4%Not measuredOnly DeepSeek V3.2 measured
CL-bench Life7.4%Not measuredOnly DeepSeek V3.2 measured
LMArena Text14251438Hy3
LMArena Text: instruction following14131426Hy3
LMArena Text: creative writing14001403Hy3
LMArena Text: English14391448Hy3
LMArena Text: non-English14091426Hy3
LMArena Text: Chinese14551485Hy3
LMArena Text: German14131449Hy3
LMArena Text: Japanese13741397Hy3
LMArena Text: Korean13701396Hy3
LMArena Text: French14291456Hy3
LMArena Text: Spanish14241457Hy3
LMArena Text: Russian14241432Hy3
LMArena Text: Polish14191447Hy3
LMArena Text: multi-turn14271436Hy3
LMArena Text: longer queries14281442Hy3
LMArena Text: writing and language14071414Hy3
LMArena Text: entertainment and media13951400Hy3
Neither model has a result on the other 23 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for DeepSeek V3.2 and Hy3
BenchmarkDeepSeek V3.2Hy3Leader
Neither model has a result on 5 boards here

What each workload costs

Three fixed token mixes at each model's listed price, so per-token rates become a bill.

Chat turn

1K input + 500 output tokens

DeepSeek V3.2$0.0005Fits in one request
Hy3$0.0004Fits in one request

Hy3 costs less

Repository review

50K input + 3K output tokens

DeepSeek V3.2$0.015Fits in one request
Hy3$0.0082Fits in one request

Hy3 costs less

Agent loop

200K input + 10K output tokens

DeepSeek V3.2$0.060Exceeds the context window
Hy3$0.032Fits in one request

Hy3 costs less

Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.

Specification differences

The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.

Specifications of DeepSeek V3.2 and Hy3
SpecificationDeepSeek V3.2Tencent Hy3
MakerDeepSeekTencent
WeightsOpen weightsOpen weights
ReleasedDecember 1, 2025July 6, 2026
Context window164K262K
Input $/Mtok$0.28$0.132
Output $/Mtok$0.42$0.528
Coding index54.443.9
Independent measurements behind it21
Cost per solved task$0.99Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is DeepSeek V3.2 better than Hy3 for coding?

Pick DeepSeek V3.2 for coding: it scores 54.4 on the Vibe Coding Index against 43.9 for Hy3. Pick Hy3 when the context window is larger, 262K against 164K tokens, or when what matters is LMArena WebDev or LMArena Text: coding, where it leads.

Which is cheaper, DeepSeek V3.2 or Hy3?

DeepSeek V3.2 costs $0.28 per million input tokens and $0.42 per million output tokens; Hy3 costs $0.132 and $0.528. Cost per solved task is known for DeepSeek V3.2 only.

Which has the bigger context window, DeepSeek V3.2 or Hy3?

Hy3, with 262K tokens against 164K.

How many benchmarks is the DeepSeek V3.2 vs Hy3 comparison based on?

42 benchmarks have a result for both models as of October 21, 2026. DeepSeek V3.2's coding index rests on 2 independent benchmarks and Hy3's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next