# Vision Medium

> Vision Medium is a proprietary language model from Vispark, released on 15 May 2024. It accepts text, images, audio, video and PDFs and generates text, with a 1M-token context window and up to 66K output tokens per response.

Source page: https://www.aimodel.directory/models/vision-medium
Last verified: 18 Nov 2026

## Specifications

- **Developer:** Vispark
- **Release date:** 15 May 2024 (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/vispark/models/vispark/vision-medium.toml)
- **Status:** Generally available (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/vispark/models/vispark/vision-medium.toml)
- **Weights:** Proprietary (API only) (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/vispark/models/vispark/vision-medium.toml)
- **Context window:** 1M (1,000,000 tokens) (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/vispark/models/vispark/vision-medium.toml)
- **Max output:** 66K tokens (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/vispark/models/vispark/vision-medium.toml)
- **Input:** Text, images, audio, video and PDFs (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/vispark/models/vispark/vision-medium.toml)
- **Output:** Text (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/vispark/models/vispark/vision-medium.toml)
- **Reasoning mode:** Yes (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/vispark/models/vispark/vision-medium.toml)
- **Tool calling:** Yes (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/vispark/models/vispark/vision-medium.toml)
- **Structured output:** Yes (source: models.dev, https://github.com/sst/models.dev/blob/dev/providers/vispark/models/vispark/vision-medium.toml)

## API pricing (USD per 1M tokens)

| Provider | Input | Output | Cached input | Context |
|---|---|---|---|---|

## FAQ

### What is the context window of Vision Medium?

Vision Medium has a context window of 1,000,000 tokens (1M) and can generate up to 65,536 tokens in a single response, according to models.dev.

### Is Vision Medium open source?

No. Vision Medium is a proprietary model; Vispark has not released its weights.

### When was Vision Medium released?

Vispark released Vision Medium on 15 May 2024, according to models.dev.

---
Data licensed CC BY 4.0. Cite as: AI Model Directory, "Vision Medium", https://www.aimodel.directory/models/vision-medium