Compare / DeepSeek-V4-Flash (0423) vs Hy3 Preview

DeepSeek-V4-Flash (0423) vs Hy3 Preview

Which is cheaper, which is more capable, and which does better at each kind of work, from 17 results on 3 sources that measured both models.

DeepSeekDeepSeek-V4-Flash (0423) Released 24 Apr 2026 · $0.050 in · $0.45 out
TencentHy3 Preview Released 23 Apr 2026 · $0.18 in · $0.60 out
DeepSeek-V4-Flash (0423) wins 161 too close to callHy3 Preview wins 0

Of the 17 results both models have, the better value on each. Results whose reported ranges overlap are too close to call.

Results as of 19 November 2026 (catalogue release 2026-11-19-3608133e8331).

DeepSeek-V4-Flash (0423) or Hy3 Preview? The short answer

They score about the same, so pick DeepSeek-V4-Flash (0423): it costs 47% less per token.

  • DeepSeek-V4-Flash (0423) is cheaper: $0.15 against $0.29 per million tokens, 47% less.
  • DeepSeek-V4-Flash (0423) scores higher on the Artificial Analysis Intelligence Index: 24.4 against 22.7.
  • DeepSeek-V4-Flash (0423) is newer: released 24 Apr 2026, against 23 Apr 2026 for Hy3 Preview.
  • On third-party benchmarks, on the results both models have: DeepSeek-V4-Flash (0423) does better for reasoning and knowledge, writing and sticking to the facts.

Head to head

24.4
Intelligence Index
22.7
$0.15
Price per million tokens, blendedlower is better
$0.29
1,048,576
Context window, tokens
262,144

Every fact side by side

DeepSeek-V4-Flash (0423)Hy3 Preview
Price per million tokens$0.050 in · $0.45 out$0.18 in · $0.60 out
Blended (3 in : 1 out)$0.15$0.29
Cheapest host, blended$0.10 (StreamLake, FP8)–
Intelligence Index24.422.7
Spring Prompt overall––
Output speed, tokens a second––
Context window1,048,576 tokens262,144 tokens
Released24 Apr 2026 (newer)23 Apr 2026
WeightsOpenOpen
Developer based inChinaChina

Prices are the developer's list price where we have read it, otherwise the typical host on OpenRouter; see every model's price. Intelligence and speed from Artificial Analysis.

Which is better for what

Each model's rank among the models on sale today, for each kind of work, from our own benchmarks and licensed sources. Further right is better; the better rank is in green.

Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls 59 of 159 not measured
Reasoning and knowledgeHard questions across science, maths and general knowledge 69 of 187 86 of 187
WritingWhat people prefer in blind comparisons, and judged writing quality 63 of 169 89 of 169
Sticking to the factsSummarising without inventing things, and factual answers 81 of 139 121 of 139

Every shared result

Each model's best result on every measure both have, grouped by benchmark, with the setting that produced it. Under each measure, every model that benchmark has measured is a grey tick, best to the right, so you can see where the two sit in the field. The headline measures of each benchmark come first; the rest are under "Show all". Green marks the better value; where a source reports ranges that overlap, neither is marked.

MeasureDeepSeek-V4-Flash (0423)Hy3 Preview
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 3 results
Overall · TextArena rating, higher is better 85 · 128 of 387 measured 1,438high reasoning 1,409
Overall · Text factualityArena rating, higher is better 81 · 125 of 163 measured 1,439 1,412
Business, management and finance · TextArena rating, higher is better 91 · 123 of 380 measured 1,439 1,411
Artificial Analysis reported by Artificial Analysis · 3 results
Artificial Analysis Intelligence Indexindex score, higher is better 85 · 95 of 249 measured 24.4high reasoning 22.7thinking reasoning
GPQA Diamond% of questions, higher is better 52 · 69 of 232 measured 89.4%max reasoning 86.7%thinking reasoning
Humanity's Last Exam% of questions, higher is better 67 · 92 of 250 measured 34.8%max reasoning 27.8%thinking reasoning
UGI Leaderboard reported by UGI Leaderboard · 3 results
Writing score · UGIscore out of 100, higher is better 60 · 76 of 230 measured 54.6no reasoning 51.1thinking reasoning
Requested-length error · UGI% off the requested word count, lower is better 54 · 91 of 230 measured 12.0%thinking reasoning 18.0%thinking reasoning
Style adherence · UGIscore from 0 to 1, higher is better 55 · 117 of 230 measured 0.37thinking reasoning 0.35no reasoning
Show all 17 results (8 more)
MeasureDeepSeek-V4-Flash (0423)Hy3 Preview
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 4 results
Creative writing · TextArena rating, higher is better 78 · 149 of 385 measured 1,410high reasoning 1,353
Expert prompts · TextArena rating, higher is better 83 · 105 of 338 measured 1,466high reasoning 1,446
Instruction following · TextArena rating, higher is better 79 · 131 of 387 measured 1,433high reasoning 1,396
Writing, literature and language · TextArena rating, higher is better 78 · 140 of 386 measured 1,422high reasoning 1,377
Artificial Analysis reported by Artificial Analysis · 4 results
IFBench% of instructions, higher is better 7 · 74 of 195 measured 79.2%max reasoning 63.1%thinking reasoning
Long-context reasoning (AA-LCR)% of questions, higher is better 89 · 136 of 241 measured 74.3%max reasoning 64.7%thinking reasoning
Terminal-Bench Hard% of tasks, higher is better 37 · 56 of 190 measured 38.6%high reasoning 34.1%thinking reasoning
τ²-bench telecom% of tasks, higher is better 16 · 35 of 193 measured 95.6%high reasoning 92.7%thinking reasoning

Quick answers

Is DeepSeek-V4-Flash (0423) better than Hy3 Preview?

They score about the same, so pick DeepSeek-V4-Flash (0423): it costs 47% less per token.

Which is cheaper, DeepSeek-V4-Flash (0423) or Hy3 Preview?

DeepSeek-V4-Flash (0423) is cheaper: $0.15 against $0.29 per million tokens, 47% less.

Which is better for reasoning and knowledge, DeepSeek-V4-Flash (0423) or Hy3 Preview?

DeepSeek-V4-Flash (0423) ranks 69 of 187 models on sale for hard questions across science, maths and general knowledge, against 86 for Hy3 Preview.

Which is better for writing, DeepSeek-V4-Flash (0423) or Hy3 Preview?

DeepSeek-V4-Flash (0423) ranks 63 of 169 models on sale for what people prefer in blind comparisons, and judged writing quality, against 89 for Hy3 Preview.

Which is better for sticking to the facts, DeepSeek-V4-Flash (0423) or Hy3 Preview?

DeepSeek-V4-Flash (0423) ranks 81 of 139 models on sale for summarising without inventing things, and factual answers, against 121 for Hy3 Preview.

Which has the bigger context window, DeepSeek-V4-Flash (0423) or Hy3 Preview?

DeepSeek-V4-Flash (0423): 1,048,576 tokens, against 262,144 for Hy3 Preview.

Run DeepSeek-V4-Flash (0423) and Hy3 Preview on your own prompt.

Benchmarks aren't your task. Test both side by side on your prompt, with every answer checked against your rules and the cost of each one.