Models / DeepSeek-V3.2

DeepSeek

DeepSeek-V3.2

DeepSeek-V3.2 is strongest at writing; worth a look for simple, high-volume work where cost matters most. The cheapest model of similar intelligence is DeepSeek-V4-Flash (0423), at $0.15 per million tokens against $0.32.

Results 16 Nov 2026 · 44 results · 5 sources

Test DeepSeek-V3.2 on your prompt → It runs beside the models you could use instead, with every answer checked and costed.

Price per million tokens
$0.28 in · $0.42 out
Typical host on OpenRouter, 10 Nov 2026 · compare prices
Speed
Not measured
Intelligence Index
21.5
Artificial Analysis, at thinking reasoning; 16.0–21.5 across 2 settings
Spring Prompt overall
Not ranked yet
needs our benchmarks and two groups of results
Developer
DeepSeek, based in China
Weights published
Released
1 Dec 2025
163,840 tokens of context

Where it stands among the models you could choose

Available models grouped by price, intelligence, speed and developer location, plus models with published weights. Retired models are excluded.

AmongIntelligence IndexSpring Prompt overallPrice, blendedSpeed
Models available today 93rd of 185Claude Opus 5.5 – $0.32 blended, 28th percentilecheapest: Mistral Nemo –
Models priced $0.25–1 per million tokens 21st of 61MiMo-V2.6-Pro – sets this group –
Models with similar intelligence sets this group – $0.32 blended, 12th percentilecheapest: DeepSeek-V4-Flash (0423) –
Models from developers based in China 38th of 75MiMo-V2.6-Pro – $0.32 blended, 26th percentilecheapest: Qwen3.5-9B –
Open-weights models 38th of 104MiMo-V2.6-Pro – $0.32 blended, 40th percentilecheapest: Mistral Nemo –

The name under each rank is the leader of that group; green is the top quarter of the group and red the bottom quarter, by rank. Price is shown as a percentile: the share of the group that costs less, so lower is cheaper. "Sets this group" marks the measure the group is defined by. Spring Prompt overall is our score out of 100 across every source (how it works). Price is the developer's list price (or the typical OpenRouter host where we haven't read one) for three input tokens to one output; intelligence and speed are from Artificial Analysis (data sourced from Artificial Analysis); "similar intelligence" means within 4 points on its Intelligence Index. Developer locations are where each company is based, not where a model is served. Groups under 3 models are not ranked.

How good is it, and for what?

Each use case ranks the available models its benchmarks measured, then those priced $0.25–1 per million tokens. The bar shows its percentile in that field, best to the right. Open a row for the results behind it.

Product listingsTurning a sparse product feed and photos into listings that can go live Not measured
  • CatalogBench: has not measured this model.
Decks from an analysisTurning a finished analysis into a deck you could present as it is Not measured
  • DeckBench: has not measured this model.
User surveysPlanning a user survey and reading its results without being misled Not measured
  • SurveyBench: has not measured this model.
Marketing planningPlanning a year of ad spend without overspending Not measured
  • ROASBench: has not measured this model.
Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls 93rd of 159 available · 21st of 47 at its price
BenchmarkDeepSeek-V3.2Rank among availableBest available
OpenHands IndexOpenHands Index: average · reported 45.7% 20th of 31 Claude Fable 5 81.0%
Artificial AnalysisArtificial Analysis: Terminal-Bench 2.1 · reported 46.8% 62nd of 107 Claude Fable 5.1 91.4%
  • tau2-bench: has not measured this model.
  • Berkeley Function Calling Leaderboard (BFCL) V4: has not measured this model.
  • Microsoft STATE-Bench: has not measured this model.
  • Vending-Bench 2: has not measured this model.
Professional workReal tasks from banking, consulting and law, business documents and freelance projects Not measured
  • APEX-Agents: has not measured this model.
  • GDP.pdf: has not measured this model.
  • Remote Labor Index: has not measured this model.
Reasoning and knowledgeHard questions across science, maths and general knowledge 93rd of 187 available · 21st of 61 at its price
BenchmarkDeepSeek-V3.2Rank among availableBest available
Artificial AnalysisArtificial Analysis: Artificial Analysis Intelligence Index · reported 21.5 93rd of 185 Claude Opus 5.5 57.6
Artificial AnalysisArtificial Analysis: Humanity's Last Exam · reported 24.6% 91st of 186 Claude Opus 5.5 61.4%
Artificial AnalysisArtificial Analysis: GPQA Diamond · reported 84.0% 78th of 168 GPT-6 Astra 96.3%
WritingWhat people prefer in blind comparisons, and judged writing quality 65th of 169 available · 10th of 58 at its price
BenchmarkDeepSeek-V3.2Rank among availableBest available
Arena (formerly LMArena)Arena Text: overall · reported 1,425 76th of 153 Gemini 4 Argon 1,525
UGI LeaderboardUGI: writing score · reported 57.3 40th of 124 Gemini 3.8 Flash 78.6
Sticking to the factsSummarising without inventing things, and factual answers 58th of 139 available · 8th of 40 at its price
BenchmarkDeepSeek-V3.2Rank among availableBest available
Vectara Hallucination LeaderboardVectara: hallucination rate · reported 6.3% 18th of 71 GPT-5.4 nano 3.1%
Arena (formerly LMArena)Arena Text factuality: overall · reported 1,437 67th of 109 Gemini 4 Argon 1,511
  • SimpleQA Verified (Epoch AI): has not measured this model.
Speed under pressureGood decisions against a real clock (fast chess) Not measured
  • BulletBench: has not measured this model.

Nearest alternatives

Against DeepSeek-V3.2 at $0.32 blended, Intelligence Index 21.5. "Similar intelligence" means within 4 points or better.

Where to run it

7 hosts on OpenRouter at the standard service tier, with a median of $0.32 blended per million tokens; quantisation where the host states it: FP8, FP4. Cheapest first.

HostQuantisationInputOutputBlended
GMICloudFP8$0.21$0.31$0.23
AtlasCloudFP8$0.26$0.38$0.29
DeepInfraFP4$0.26$0.38$0.29
BaiduFP8$0.28$0.42$0.32
SiliconFlowFP8$0.26$0.80$0.39
DigitalOceannot stated$0.30$0.96$0.46
Phalanot stated$1.00$1.00$1.00

Prices per million tokens from OpenRouter's public endpoint list, checked 10 Nov 2026.

Against the alternatives

The models you would most likely weigh it against: the leaders of the groups above.

Scroll sideways to see every alternative.

DeepSeek-V3.2MiMo-V2.6-Pro
most intelligent at $0.25–1 per million tokens
DeepSeek-V4-Flash (0423)
cheapest with similar intelligence
Claude Opus 5.5
most intelligent available today
Price per million tokens $0.32 $0.54$0.15$8.00
Tokens a second – 43–96
Intelligence Index 21.5 46.324.457.6
Spring Prompt overall – ––82
Rank among available models, by use case
Product listings – ––13th
Decks from an analysis – ––7th
User surveys – ––4th
Marketing planning – ––9th
Agents and tool use 93rd of 159 23rd59th3rd
Professional work – ––3rd
Reasoning and knowledge 93rd of 187 13th69th1st
Writing 65th of 169 23rd63rd5th
Sticking to the facts 58th of 139 10th81st3rd
Speed under pressure – –––

Shaded figures are better than DeepSeek-V3.2's. A dash means no result.

Where it has been measured

21st most intelligent of the 61 models priced $0.25–1 per million tokens. Among models available today, mid-field on every use case measured so far.

Artificial Analysis lists it but has not published a speed for it yet

Every result

44 published results from 5 sources, each in the source's own units, with the configuration that produced it.

Arena (formerly LMArena) · reported by Arena (formerly LMArena) · 14 results
MeasureValueRankConfigurationDated
Arena Text factuality: overallArena rating, higher is better 1,437
1,434–1,440
98 of 182 DeepSeek-V3.2 2 Oct 2026
Arena Text factuality: overallArena rating, higher is better 1,433
1,429–1,436
110 of 182 DeepSeek-V3.2 (thinking reasoning) 2 Oct 2026
Arena Text: business, management and financeArena rating, higher is better 1,422
1,415–1,429
121 of 406 DeepSeek-V3.2 2 Oct 2026
Arena Text: business, management and financeArena rating, higher is better 1,423
1,416–1,430
118 of 406 DeepSeek-V3.2 (thinking reasoning) 2 Oct 2026
Arena Text: creative writingArena rating, higher is better 1,401
1,393–1,408
108 of 411 DeepSeek-V3.2 2 Oct 2026
Arena Text: creative writingArena rating, higher is better 1,392
1,384–1,399
121 of 411 DeepSeek-V3.2 (thinking reasoning) 2 Oct 2026
Arena Text: expert promptsArena rating, higher is better 1,448
1,438–1,458
118 of 363 DeepSeek-V3.2 2 Oct 2026
Arena Text: expert promptsArena rating, higher is better 1,452
1,441–1,464
109 of 363 DeepSeek-V3.2 (thinking reasoning) 2 Oct 2026
Arena Text: instruction followingArena rating, higher is better 1,420
1,414–1,426
110 of 413 DeepSeek-V3.2 2 Oct 2026
Arena Text: instruction followingArena rating, higher is better 1,420
1,414–1,425
111 of 413 DeepSeek-V3.2 (thinking reasoning) 2 Oct 2026
Arena Text: overallArena rating, higher is better 1,425
1,421–1,428
122 of 413 DeepSeek-V3.2 2 Oct 2026
Arena Text: overallArena rating, higher is better 1,423
1,419–1,426
125 of 413 DeepSeek-V3.2 (thinking reasoning) 2 Oct 2026
Arena Text: writing, literature and languageArena rating, higher is better 1,410
1,403–1,416
106 of 412 DeepSeek-V3.2 2 Oct 2026
Arena Text: writing, literature and languageArena rating, higher is better 1,404
1,397–1,410
121 of 412 DeepSeek-V3.2 (thinking reasoning) 2 Oct 2026

Not shown: Not an error rate: claims that cannot be checked on the web are skipped, and preference still carries most of the weight.

Artificial Analysis · reported by Artificial Analysis · 16 results
MeasureValueRankConfigurationDated
Artificial Analysis: Artificial Analysis Coding Indexindex score, higher is better 44.2 112 of 187 DeepSeek-V3.2 (thinking reasoning) 20 Oct 2026
Artificial Analysis: Artificial Analysis Intelligence Indexindex score, higher is better 16.0 245 of 413 DeepSeek-V3.2 (no reasoning) 20 Oct 2026
Artificial Analysis: Artificial Analysis Intelligence Indexindex score, higher is better 21.5 193 of 413 DeepSeek-V3.2 (thinking reasoning) 20 Oct 2026
Artificial Analysis: GPQA Diamond% of questions, higher is better 75.1% 217 of 360 DeepSeek-V3.2 (no reasoning) 20 Oct 2026
Artificial Analysis: GPQA Diamond% of questions, higher is better 84.0% 139 of 360 DeepSeek-V3.2 (thinking reasoning) 20 Oct 2026
Artificial Analysis: Humanity's Last Exam% of questions, higher is better 11.2% 267 of 419 DeepSeek-V3.2 (no reasoning) 20 Oct 2026
Artificial Analysis: Humanity's Last Exam% of questions, higher is better 24.6% 180 of 419 DeepSeek-V3.2 (thinking reasoning) 20 Oct 2026
Artificial Analysis: IFBench% of instructions, higher is better 49.0% 145 of 288 DeepSeek-V3.2 (no reasoning) 20 Oct 2026
Artificial Analysis: IFBench% of instructions, higher is better 60.7% 104 of 288 DeepSeek-V3.2 (thinking reasoning) 20 Oct 2026
Artificial Analysis: Long-context reasoning (AA-LCR)% of questions, higher is better 45.7% 293 of 407 DeepSeek-V3.2 (no reasoning) 20 Oct 2026
Artificial Analysis: Long-context reasoning (AA-LCR)% of questions, higher is better 73.3% 171 of 407 DeepSeek-V3.2 (thinking reasoning) 20 Oct 2026
Artificial Analysis: Terminal-Bench 2.1% of tasks, higher is better 46.8% 113 of 188 DeepSeek-V3.2 (thinking reasoning) 20 Oct 2026
Artificial Analysis: Terminal-Bench Hard% of tasks, higher is better 32.6% 93 of 282 DeepSeek-V3.2 (no reasoning) 20 Oct 2026
Artificial Analysis: Terminal-Bench Hard% of tasks, higher is better 35.6% 71 of 282 DeepSeek-V3.2 (thinking reasoning) 20 Oct 2026
Artificial Analysis: τ²-bench telecom% of tasks, higher is better 78.9% 105 of 286 DeepSeek-V3.2 (no reasoning) 20 Oct 2026
Artificial Analysis: τ²-bench telecom% of tasks, higher is better 90.6% 54 of 286 DeepSeek-V3.2 (thinking reasoning) 20 Oct 2026

Not shown: Not your codebase or tools.

OpenHands Index · reported by OpenHands Index · 6 results
MeasureValueRankConfigurationDated
OpenHands Index: average% resolved, higher is better 45.7% 21 of 34 DeepSeek-V3.2 (thinking reasoning, openhands) 30 Jun 2026
OpenHands Index: front end (SWE-Bench Multimodal)% resolved, higher is better 27.9% 24 of 34 DeepSeek-V3.2 (thinking reasoning, openhands) 30 Jun 2026
OpenHands Index: greenfield (Commit0)% resolved, higher is better 25.0% 14 of 34 DeepSeek-V3.2 (thinking reasoning, openhands) 30 Jun 2026
OpenHands Index: information gathering (GAIA)% resolved, higher is better 50.3% 22 of 34 DeepSeek-V3.2 (thinking reasoning, openhands) 30 Jun 2026
OpenHands Index: issue resolution (SWE-Bench)% resolved, higher is better 71.6% 24 of 34 DeepSeek-V3.2 (thinking reasoning, openhands) 30 Jun 2026
OpenHands Index: testing (SWT-Bench)% resolved, higher is better 53.6% 25 of 34 DeepSeek-V3.2 (thinking reasoning, openhands) 30 Jun 2026

Not shown: Not the model alone: the score is for the model inside the OpenHands agent, at the SDK version shown.

UGI Leaderboard · reported by UGI Leaderboard · 6 results
MeasureValueRankConfigurationDated
UGI: requested-length error% off the requested word count, lower is better 26.0% 241 of 379 DeepSeek-V3.2 (no reasoning) 3 Dec 2025
UGI: requested-length error% off the requested word count, lower is better 29.0% 257 of 379 DeepSeek-V3.2 (thinking reasoning) 3 Dec 2025
UGI: style adherencescore from 0 to 1, higher is better 0.34 241 of 379 DeepSeek-V3.2 (no reasoning) 3 Dec 2025
UGI: style adherencescore from 0 to 1, higher is better 0.37 120 of 379 DeepSeek-V3.2 (thinking reasoning) 3 Dec 2025
UGI: writing scorescore out of 100, higher is better 56.3 128 of 379 DeepSeek-V3.2 (no reasoning) 3 Dec 2025
UGI: writing scorescore out of 100, higher is better 57.3 121 of 379 DeepSeek-V3.2 (thinking reasoning) 3 Dec 2025

Not shown: Not other format limits such as character counts or bullet counts.

Vectara Hallucination Leaderboard · reported by Vectara Hallucination Leaderboard · 2 results
MeasureValueRankConfigurationDated
Vectara: answer rate% of documents, higher is better 92.6% 98 of 108 DeepSeek-V3.2 22 Sep 2026
Vectara: hallucination rate% of summaries, lower is better 6.3% 27 of 108 DeepSeek-V3.2 22 Sep 2026

Not shown: Not a quality score: a low rate usually means content filters were triggered, and hallucination rates are measured on answered documents only.

Will DeepSeek-V3.2 work for your product?

Benchmarks aren't your task. Paste the prompt your product uses and run it beside the models you could use instead, with every answer checked and costed.

New here? Your email starts a free trial. No credit card needed.