DataLearner logo

DeepSeek-V4-FlashvsHy3 Pre

DeepSeek-V4-Flash and Hy3 Pre are tied across 5 shared benchmarks: DeepSeek-V4-Flash leads on 2, Hy3 Pre leads on 2, with 1 ties and an average score difference of +5.28.

DeepSeek-AI
DeepSeek-V4-Flash

DeepSeek-AI · 2026-04-24 · Reasoning model

腾讯AI实验室
Hy3 Pre

腾讯AI实验室 · 2026-04-23 · Reasoning model

DeepSeek-V4-Flash2 wins(40%)Ties1(40%)2 winsHy3 Pre

Benchmark scores

Grouped by capability, sorted by largest gap within each. 5 shared benchmarks.

Agent Level Benchmark

DeepSeek-V4-Flash 2/2
BenchmarkDeepSeek-V4-FlashHy3 PreDiff
τ²-Bench - Telecom94.4035 / 264Normal (With Tools)67.50143 / 264Normal (With Tools)+26.90
Terminal Bench Hard34.1082 / 244Normal (With Tools)31.80100 / 244Normal (With Tools)+2.30

General Evaluation

Hy3 Pre 1/1
BenchmarkDeepSeek-V4-FlashHy3 PreDiff
GPQA Diamond71.20311 / 462Normal (No Tools)73.20299 / 462Normal (No Tools)-2

General Knowledge

Even 1/1
BenchmarkDeepSeek-V4-FlashHy3 PreDiff
CritPt0.30182 / 200Normal (No Tools)0.30182 / 200Normal (No Tools)

Instruction Following

Hy3 Pre 1/1
BenchmarkDeepSeek-V4-FlashHy3 PreDiff
IF Bench47.20170 / 282Normal (No Tools)48166 / 282Normal (No Tools)-0.80

Specs

FieldDeepSeek-V4-FlashHy3 Pre
PublisherDeepSeek-AI腾讯AI实验室
Release date2026-04-242026-04-23
Model typeReasoning modelReasoning model
ArchitectureMoEMoE
Parameters284B295B
Context length1M256K
Max output384KNot available

API pricing

Prices use DataLearner records when available; missing fields are not inferred.

ItemDeepSeek-V4-FlashHy3 Pre
Text input$0.14 / 1M tokensNot public
Text output$0.28 / 1M tokensNot public
Cache read$0.0028 / 1M tokensNot public

One or both models have incomplete public pricing.

Summary

  • DeepSeek-V4-Flashleads in:Agent Level Benchmark (2/2)
  • Hy3 Preleads in:General Evaluation (1/1), Instruction Following (1/1)
  • Tied in:General Knowledge

On average across the 5 shared benchmarks, DeepSeek-V4-Flash scores 5.28 higher.

Largest single-benchmark gap: τ²-Bench - Telecom — DeepSeek-V4-Flash 94.40 vs Hy3 Pre 67.50 (+26.90).

Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.