# DeepSeek V3.2 Speciale vs GPT OSS 20B

Canonical HTML page: [DeepSeek V3.2 Speciale vs GPT OSS 20B](<https://impossibl.com/compare/deepseek/deepseek-v3-2-speciale/openai/gpt-oss-20b>)
Markdown alternative: [DeepSeek V3.2 Speciale vs GPT OSS 20B as Markdown](<https://impossibl.com/compare/deepseek/deepseek-v3-2-speciale/openai/gpt-oss-20b.md>)

Compare API pricing, context windows, and capabilities side by side. Prices are published API usage rates in USD; check billing units and context tiers alongside each price.

## Specifications and pricing

| Specification | DeepSeek V3.2 Speciale | GPT OSS 20B |
| --- | --- | --- |
| API identifier | deepseek/deepseek-v3-2-speciale | openai/gpt-oss-20b |
| Creator | DeepSeek | OpenAI |
| Model type | Text generation | Text generation |
| Inputs | text | text |
| Outputs | text | text |
| Base input price (USD) | $0.58 per million input tokens | $0.075 per million input tokens |
| Base output price (USD) | $1.68 per million output tokens | $0.30 per million output tokens |
| Cache read price (USD) | Not published | $0.0375 per million cached input tokens |
| Cache write price (USD) | Not published | Not published |
| One-hour cache write price (USD) | Not published | Not published |
| Audio input token price (USD) | Not published | Not published |
| Context window | 163,840 tokens | 128,000 tokens |
| Tool calling | Not published | Not published |
| Endpoints | /v1/chat/completions | /v1/chat/completions |
| Context pricing tiers (USD per million tokens) | None published | None published |
| Retirement | Not announced in the public catalog | Not announced in the public catalog |

When total input, including cached input, strictly exceeds a context threshold, the highest qualifying tier applies to the whole request. Consult each model's complete Markdown for tier-specific cache rates; an unpublished tier cache price does not establish the base cache rate.

## DeepSeek V3.2 Speciale vs GPT OSS 20B: side-by-side summary

DeepSeek V3.2 Speciale and GPT OSS 20B are available through the Impossibl API.

They share the `/v1/chat/completions` endpoint. Set the model identifier to select the model for a supported request.

DeepSeek V3.2 Speciale is created by DeepSeek. It has a 163,840-token context window. Base API pricing (USD): $0.58 per million input tokens; $1.68 per million output tokens.

GPT OSS 20B is created by OpenAI. It has a 128,000-token context window. Base API pricing (USD): $0.075 per million input tokens; $0.30 per million output tokens.

## Model details

- [DeepSeek V3.2 Speciale: full model facts as Markdown](<https://impossibl.com/deepseek/deepseek-v3-2-speciale.md>); [HTML model page](<https://impossibl.com/deepseek/deepseek-v3-2-speciale>)
- [GPT OSS 20B: full model facts as Markdown](<https://impossibl.com/openai/gpt-oss-20b.md>); [HTML model page](<https://impossibl.com/openai/gpt-oss-20b>)

For compatible text or embedding models with token pricing, the HTML comparison page provides a shared workload calculator. It excludes cache creation, tool calls, taxes, balance-purchase fees, and workspace-specific adjustments.

## Sources and related links

Prices, capabilities, limits, and availability describe the public Impossibl API offering. An unpublished fact is unknown; a published zero price is zero. Compare matching billing units and check context tiers before estimating costs.

- [Public Impossibl model catalog](<https://api.impossibl.com/v1/models>)
- [Full model directory as JSON](<https://impossibl.com/models.json>)
- [Model directory as Markdown](<https://impossibl.com/models.md>)
- [API quickstart](<https://impossibl.com/docs/agent-quickstart>)
- [Model documentation](<https://impossibl.com/docs/models>)