DeepSeek V3.2 Speciale
29 cheaper ranked models score higher: DeepSeek V4.1 Flash 1017 (59.2, $0.525), DeepSeek V4.1 Flash (59.0, $0.518), GPT-5.6 Luna (58.5, $0.450) and 26 more.
As of 8 Nov 2026, DeepSeek V3.2 Speciale by DeepSeek scores 53.7 on the BenchLeader Index (593 ranked configurations), with its best results in agents & tools (64) and weakest in maths (45). API pricing: $0.580 per 1M input tokens and $1.68 per 1M output tokens ($0.855 blended at 3:1), list price on 8 Nov 2026. Released 1 Dec 2025.
Key facts (as of 8 Nov 2026)
- DeepSeek V3.2 Speciale API pricing: $0.580 per 1M input tokens and $1.68 per 1M output tokens ($0.855 blended at 3:1) (list price, 8 Nov 2026).
- DeepSeek V3.2 Speciale was released on 1 Dec 2025 (DeepSeek).
In detail. DeepSeek V3.2 Speciale is a DeepSeek open-weights reasoning model, released 1 Dec 2025. Its best configuration ranks #247 of 593 on the BenchLeader Index at 53.7 ±6.0, in the upper half. The ± is the point: 323 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in agents & tools (64) and lowest in maths (45). At $0.855 per million tokens blended it is mid-priced. Last measured 1 Dec 2025.stale: no new result in six months
- Blended price
- $0.855/M
- $0.580 in · $1.68 out
- via models.dev
- Output speed
- –
- First answer
- –
- Context
- 128k
- via Artificial Analysis
- Full answer
- –
- Cost per task
- –
- Tokens per task
- –
- Released
- 1 Dec 2025
- Overall index54
- Reasoning53
- Coding48
- Agents & tools64
- Maths45
- Instruction following64
- Long context60
How its index has moved
9 Sep 2026 to 8 Nov 2026The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | not stated | Source |
|---|---|---|
| SimpleBench | 52.6%#43 | SimpleBench |
| GPQA Diamond (AA)not in index | 87.1%#105 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 28.7%#168 | Artificial Analysis |
| CritPtnot in index | 7.4%#149 | Artificial Analysis |
Coding
| Benchmark | not stated | Source |
|---|---|---|
| WeirdML | 46.7%#77 | WeirdML |
| LiveCodeBench (AA)not in index | 89.6%#3 | Artificial Analysis |
Agents & tools
| Benchmark | not stated | Source |
|---|---|---|
| Terminal-Bench Hard | 34.9%#82 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 0.0%#398 | Artificial Analysis |
Maths
| Benchmark | not stated | Source |
|---|---|---|
| MathArena Apexpossible contamination | 9.4%#22 | MathArena |
Knowledge
| Benchmark | not stated | Source |
|---|---|---|
| Global-MMLU-Lite (AA)not in index | 88.4%#19 | Artificial Analysis |
Instruction following
| Benchmark | not stated | Source |
|---|---|---|
| IFBench | 63.9%#113 | Artificial Analysis |
Long context
| Benchmark | not stated | Source |
|---|---|---|
| AA-LCR v1.1 | 70.0%#224 | Artificial Analysis |
Composite
| Benchmark | not stated | Source |
|---|---|---|
| AA Intelligence Index v4.3.2not in index | 14.5#304 | Artificial Analysis |
| AA Progress Index v1not in index | 264.5#227 | Artificial Analysis |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0007 | – | – |
| Summarize a 30-page report | 12,000 / 600 | $0.0080 | – | – |
| Code edit | 6,000 / 1,500 | $0.0060 | – | – |
| Agentic coding session | 60,000 / 4,000 | $0.042 | – | – |
| Structured extraction | 2,000 / 200 | $0.0015 | – | – |
More from DeepSeek
See also
Embed or cite
Show where DeepSeek V3.2 Speciale ranks on a model card or README. The badge redraws itself with every daily update, carries the data date and links back to this page. To see ranks like this on OpenRouter and Hugging Face model pages, get the Chrome extension.
[](https://www.benchleader.com/models/deepseek-v3-2-speciale)Cite this
The ranking changes daily, so a reference carries the data date. APA style:
BenchLeader. (2026, November 8). DeepSeek V3.2 Speciale on the BenchLeader Index (method v1.4) [Data set]. Retrieved November 8, 2026, from https://www.benchleader.com/models/deepseek-v3-2-speciale
BibTeX:
@misc{benchleader_deepseek_v3_2_speciale_20261108,
author = {{BenchLeader}},
title = {{DeepSeek V3.2 Speciale on the BenchLeader Index (method v1.4)}},
year = {2026},
month = nov,
howpublished = {\url{https://www.benchleader.com/models/deepseek-v3-2-speciale}},
note = {Data as of 2026-11-08}
}Every badge style, the embeddable top-10 widget and how to cite the Index
Questions people ask
- What are DeepSeek V3.2 Speciale's benchmark scores?
- As of 8 Nov 2026, DeepSeek V3.2 Speciale scores 53.7 ±6.0 on the BenchLeader Index, strongest in agents & tools (64) and weakest in maths (45). That rests on independent benchmark results only, and every score on this page links to its publisher.
- How does DeepSeek V3.2 Speciale rank?
- DeepSeek V3.2 Speciale is #247 of 593 ranked configurations on 8 Nov 2026. 323 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 69.4.
- How much does DeepSeek V3.2 Speciale cost per 1M tokens?
- As of 8 Nov 2026, DeepSeek V3.2 Speciale costs $0.580 per 1M input tokens and $1.68 per 1M output tokens, $0.855 blended (three input tokens per output token).
- When was DeepSeek V3.2 Speciale released?
- DeepSeek released DeepSeek V3.2 Speciale on 1 Dec 2025.
Data as of 8 Nov 2026. Compare with another model.
Cite as: BenchLeader, “DeepSeek V3.2 Speciale: benchmarks, pricing, speed and rank”, data as of 8 Nov 2026, https://www.benchleader.com/models/deepseek-v3-2-speciale.