DeepSeek · Released Dec 1, 2025DeepSeek V3.2
Hybrid-reasoning DeepSeek model with thinking and non-thinking modes, sparse attention, and tool-use.
- Capability
- 146.3ECI · #69 of 148
- Input
- $0.296per 1M tokens
- Output
- $0.48per 1M tokens
- Context
- 128K64K max output
What it costs
- Input
- $0.296per million tokens
- Output
- $0.48per million tokens
- Blended (3:1)
- $0.342cheaper than 69% of priced models
Typical monthly bills
| Monthly usage | Estimated cost |
|---|---|
| Side project2M input + 0.5M output tokens | $0.832 |
| Team assistant25M input + 5M output tokens | $9.80 |
| Production app250M input + 50M output tokens | $98.00 |
Median of 15 paid API providers on models.dev.
Independent benchmarks
#69 of 148 scored models
The shaded band is Epoch AI’s confidence range (144.4–147.5); the tick marks the median scored model.
Scores from Epoch AI, run independently of DeepSeek.
The details
- Lab
- DeepSeek
- Released
- Dec 1, 2025
- Knowledge cutoff
- Jul 2024
- Context window
- 128,000 tokens
- Max output
- 64,000 tokens
- Inputs
- Text
- Output
- Text
- Reasoning
- Always on
- Tool calling
- Yes
- Structured output
- Yes
- Weights
- Open weights · MIT License
- Availability
- 15 API providerslisted on models.dev
More from DeepSeek
About DeepSeek V3.2
How much does DeepSeek V3.2 cost?
DeepSeek V3.2 costs $0.296 input / $0.48 output per million tokens (median across 15 API providers). At a 3:1 input-to-output mix that is $0.342 per million tokens, cheaper than 69% of the 360 priced models we track.
What is the context window of DeepSeek V3.2?
DeepSeek V3.2 accepts up to 128,000 tokens per request and can write up to 64,000 tokens in one response.
How good is DeepSeek V3.2?
Epoch AI gives DeepSeek V3.2 a Capabilities Index score of 146.3 (likely range 144.4–147.5), ranking it #69 of 148 models Epoch has scored. Epoch AI benchmark results: GPQA Diamond 83.4%, OTIS Mock AIME 2024–2025 87.8%.
Is DeepSeek V3.2 open source?
Yes. DeepSeek publishes the weights under the MIT License licence, so it can be downloaded and self-hosted.
What inputs does DeepSeek V3.2 support?
DeepSeek V3.2 accepts text and replies in text. It is a reasoning model, supports tool calling and can return structured JSON output.
When was DeepSeek V3.2 released?
DeepSeek released DeepSeek V3.2 on Dec 1, 2025. Its training data runs to Jul 2024.
What are the best alternatives to DeepSeek V3.2?
The closest current models from other labs on capability, price and release date are GPT-5.4 nano (OpenAI, ECI 145.8, $0.20 / $1.25), MiniMax-M2.5 (MiniMax, ECI 146.7, $0.30 / $1.20), Kimi K2 Thinking (Moonshot AI, ECI 146.0, $0.60 / $2.50) and Qwen3.6 Flash (Alibaba (Qwen), ECI 143.3, $0.188 / $1.13).