DeepSeek V3.2 Thinking
deepseek/v3-2-thinkingBy DeepSeek · family: deepseek · released 2025-12-01 · knowledge: 2024-12
Prices in USD per 1M tokens. Unknown means the provider does not publish per-token pricing.
Capabilities
Model fit scores
0–100 · higher is betterThese scores reward declared capabilities, context size, price and provider availability — they are not benchmark results. Use them as a directional signal alongside your own evaluation.
Coding55
- Tool calling40/40
- Structured output0/20
- Reasoning10/10
- Context window (100K → 1M)2/20
- Provider availability3/10
Agents68
- Tool calling35/35
- Structured output0/25
- Reasoning15/15
- Output token limit15/15
- Provider availability3/10
JSON / structured output49
- Structured output / JSON mode0/50
- Tool calling20/20
- Temperature control10/10
- Price-friendly for high-volume19/20
Cost efficiency71
- Headline price (log-scaled)66/95
- Has prompt-cache pricing5/5
Long context45
- Context window (100K → 2M)35/90
- Has published price for full window10/10
Production-readiness71
- Number of independent providers15/40
- Has published per-token price20/20
- Context window ≥ 8K15/15
- No data inconsistencies across providers6/10
- Official model (not derivative)15/15
Cost Efficiency Index
Open full calculator →Estimated cost using the recommended provider's headline rate. Each scenario fixes average input/output tokens — the assumptions are shown in the third column.
| Scenario | Cost | Assumption |
|---|---|---|
RAG answer per 1,000 RAG answers | $1.61 < $0.01 per request | 5K input tokens (query + 4 retrieved chunks of ~1K each) and a 500-token answer. Typical SaaS knowledge-base bot. |
Support ticket triage per 10,000 tickets | $3.22 < $0.01 per request | 1K input tokens (ticket body + system prompt) and a 100-token JSON classification reply. High-volume customer support. |
Data extraction per 1,000 documents | $0.77 < $0.01 per request | 2K input tokens (a single document page) and a 500-token JSON extraction. ETL / invoice / form pipelines. |
Code review per 1,000 PRs | $2.66 < $0.01 per request | 8K input tokens (diff + surrounding files) and a 1K-token review comment. PR-bot workloads. |
Agent step per 1,000 steps | $3.61 < $0.01 per request | 12K input tokens (long-running tool history) and a 600-token tool-call decision. Cost per agent step. |
Pricing detail
Recommended pricing from vercel · deepseek/deepseek-v3.2-thinking
Available on 3 providers
| Provider | Provider model id | Input / 1M | Output / 1M | Context | Released |
|---|---|---|---|---|---|
| Vercel AI Gateway vercel | deepseek/deepseek-v3.2-thinking | $0.280 | $0.420 | 128K | 2025-12-01 |
| 302.AI 302ai | deepseek-v3.2-thinking | $0.290 | $0.430 | 128K | 2025-12-01 |
| NanoGPT nano-gpt | deepseek/deepseek-v3.2:thinking | $0.280 | $0.420 | 163K | 2025-12-01 |
Data inconsistencies across providers
- context_window varies: 128000, 163000
- modalities varies across offerings
Different providers report different values for this model. Quick facts above use the representative provider; consult the table for per-provider truth.
Frequently asked questions
How much does DeepSeek V3.2 Thinking cost?
DeepSeek V3.2 Thinking costs $0.280 per 1M input tokens and $0.420 per 1M output tokens, sourced from vercel. Cache reads, audio tokens and >200K-context tiers (where applicable) are listed in the Pricing detail block above.
What is the context window of DeepSeek V3.2 Thinking?
DeepSeek V3.2 Thinking has a context window of 128K tokens, with a max output of 64K tokens per reply. This is the total combined size of prompt + completion.
Does DeepSeek V3.2 Thinking support tool calling?
Yes. DeepSeek V3.2 Thinking supports tool calling (function calling). This makes it suitable for production agent and automation workloads where the model has to invoke external functions reliably.
Does DeepSeek V3.2 Thinking support structured output / JSON mode?
Support for structured output / JSON-schema-constrained decoding is not reported for DeepSeek V3.2 Thinking in our data source. Verify with DeepSeek's official documentation before relying on it in production.
Can DeepSeek V3.2 Thinking accept image input?
No. DeepSeek V3.2 Thinking only accepts text as input. If you need image input, see our /capabilities/vision list for current vision-capable models.
Is DeepSeek V3.2 Thinking open-weight?
No. DeepSeek V3.2 Thinking is a proprietary model — only DeepSeek (and any approved hosting partners) can serve it. The pricing above reflects the cheapest API access.
What are the best alternatives to DeepSeek V3.2 Thinking?
If DeepSeek V3.2 Thinking doesn't fit, consider DeepSeek-V3.2, DeepSeek V4 Pro, DeepSeek-R1-0528. Each one targets the same use case — see the Related links below for direct head-to-head pages.
Where does this data come from?
All numbers come from the public models.dev API and are normalised into a single canonical model record. We re-pull daily and write any changes (price, context, capability) to the /changelog page.
Explore more
More DeepSeek models
- DeepSeek-V3.2$0.26 in / $0.38 out
- DeepSeek V4 Pro$1.74 in / $3.48 out
- DeepSeek-R1-0528$0.40 in / $1.70 out
- DeepSeek-R1$0.40 in / $1.70 out
- DeepSeek-V3.1$0.20 in / $0.70 out
Capability lists this model is in
Last updated:
Data is sourced from models.dev and normalized for comparison. Prices and capabilities may change. Always verify critical production decisions with the provider's official documentation.