DeepSeek-V3.2-Speciale
DeepSeek-V3.2-Speciale is the API-only, maximum-reasoning variant released alongside DeepSeek-V3.2.
text + code → text + code · made by DeepSeek
$0.28→$0.42per Mtok in / out
ElectronHub · 2 sellers
Sold by 2 ways
| Seller | Lane | Rate |
|---|---|---|
| Microsoft Azure AI cloud | standard | $0.58per Mtok in$1.68per Mtok outmodels.dev · read 2026-09-16 |
| ElectronHub aggregator | standard | $0.28per Mtok in$0.42per Mtok outapi.electronhub.ai · read 2026-09-16 |
Measured 3 standings
| Place | Board | Metric | Score |
|---|---|---|---|
| 6thof 61 | AIME 2025 | Accuracy (± 95% CI) | 95.83 |
| 48thof 102 | Simplebench — Epoch AI | Score (AVG@5) | 0.526 |
| 78thof 170 | Weirdml — Epoch AI | Accuracy | 46.73 |
About
DeepSeek-V3.2-Speciale — a text and code model from DeepSeek, sold by 2 companies from $0.28 in and $0.42 out per million tokens, placed 6th of 61 on AIME 2025.
It takes text and code and returns text and code, with a context window of 128,000 tokens. It was published in December 2025, trained on material up to July 2024. Its sellers say it can reason step by step. The catalogue files it under code and reasoning. Two companies sell it. The cheapest is $0.28 in and $0.42 out per million tokens at ElectronHub. It has been measured on 3 boards, and stands best at 6th of 61 on AIME 2025.
Every current figure
- Maker
- DeepSeek
- Register
- model
- Takes
- text + code
- Returns
- text + code
- Context
- 128,000 tokens
- Published
- December 2025
- Knowledge to
- July 2024
- Licence
- not read
- Sellers
- 2
- Maker's own price
- not read
- Price
- $0.28 in and $0.42 out per million tokens — ElectronHub
- Boards
- 3
- Best place
- 6th of 61 — AIME 2025
Known as 1 name
V3.2 SP