> ## Documentation Index
> Fetch the complete documentation index at: https://docs.zerogpu.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Open Weight

> The open-weight frontier-class models served on ZeroGPU, with pricing and context windows.

Open-weight models ship with published weights and permissive licenses, so you can inspect them, self-host them, and avoid lock-in to a closed vendor. ZeroGPU serves ten of them — for reasoning, coding, agentic workflows, multimodal understanding, high-volume summarization, document OCR, tool selection, and safety classification — through the same OpenAI-compatible API as every other model in the [catalog](https://docs.zerogpu.ai/docs/model-catalog.md).

Full request and response examples live on the [Text Generation](https://docs.zerogpu.ai/docs/text-generation.md) page. Open-weight speech-to-text and text-to-speech models are on [Audio Models](https://docs.zerogpu.ai/docs/audio.md).

## At a glance

| Model | Input /1M | Output /1M | Cached input /1M | Max tokens |
| - | -: | -: | -: | -: |
| [![deepseek-v4.1-flash](https://models-favicon.zerogpu.ai/deepseek-ai/deepseek-color.png) `deepseek-v4.1-flash`](https://docs.zerogpu.ai/api-reference/models/deepseek-v4-1-flash) | \$0.30 | \$1.20 | \$0.003 | 1,048,576 |
| [![glm-5.3-flash](https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png) `glm-5.3-flash`](https://docs.zerogpu.ai/api-reference/models/glm-5-3-flash) | \$0.10 | \$0.35 | \$0.015 | 1,048,576 |
| [![qwen3-30b-a3b-fp8](https://models-favicon.zerogpu.ai/qwen3-30b-a3b-fp8/qwen3-30b-a3b-fp8.png) `qwen3-30b-a3b-fp8`](https://docs.zerogpu.ai/api-reference/models/qwen3-30b-a3b-fp8) | \$0.10 | \$0.45 | \$0.05 | 32,768 |
| [![qwen3.6-35b-a3b](https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg) `qwen3.6-35b-a3b`](https://docs.zerogpu.ai/api-reference/models/qwen3-6-35b-a3b) | \$0.20 | \$1.50 | \$0.20 | 16,384 |
| [![gpt-oss-120b](https://models-favicon.zerogpu.ai/gpt-oss-120b/gpt-oss-120b.png) `gpt-oss-120b`](https://docs.zerogpu.ai/api-reference/models/gpt-oss-120b) | \$0.15 | \$0.60 | \$0.03 | 131,072 |
| [![llama-3.1-8b-instruct-fast](https://models-favicon.zerogpu.ai/llama-3.2-3b-instruct/meta.png) `llama-3.1-8b-instruct-fast`](https://docs.zerogpu.ai/api-reference/models/llama-3-1-8b-instruct-fast) | \$0.15 | \$0.28 | \$0.025 | 131,072 |
| [![llama-3.2-3b-instruct-fast](https://models-favicon.zerogpu.ai/llama-3.2-3b-instruct/meta.png) `llama-3.2-3b-instruct-fast`](https://docs.zerogpu.ai/api-reference/models/llama-3-2-3b-instruct-fast) | \$0.05 | \$0.30 | \$0.025 | 131,072 |
| [![llama-guard-4-12b](https://models-favicon.zerogpu.ai/llama-3.2-3b-instruct/meta.png) `llama-guard-4-12b`](https://docs.zerogpu.ai/api-reference/models/llama-guard-4-12b) | \$0.18 | \$0.18 | — | 163,840 |
| [`zlm-v1-tool-decisions`](https://docs.zerogpu.ai/api-reference/models/zlm-v1-tool-decisions) | \$0.05 | \$0.15 | — | 24,576 |
| [`unlimited-ocr`](https://docs.zerogpu.ai/api-reference/models/unlimited-ocr) | Per page | Per page | — | — |

## Model cards

> [card]
>
> ![deepseek-v4.1-flash](https://models-favicon.zerogpu.ai/deepseek-ai/deepseek-color.png)
>
> deepseek-v4.1-flash
>
> 1,048,576 context window\$0.30 / 1M input\$1.20 / 1M output\$0.003 / 1M cached input
>
> DeepSeek's DeepSeek-V4.1-Flash is an open-weight sparse Mixture-of-Experts model and the first built on DeepSeek's Causal Encoder-Decoder (CED) architecture, activating 8B parameters on input and 16B on output. It…
>
> [Open](https://docs.zerogpu.ai/api-reference/models/deepseek-v4-1-flash.md)

> [card]
>
> ![glm-5.3-flash](https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png)
>
> glm-5.3-flash
>
> 1,048,576 context window\$0.10 / 1M input\$0.35 / 1M output\$0.015 / 1M cached input
>
> Z.ai's GLM-5.3-Flash is an efficient open-weight model for coding and long-horizon agent tasks, served on ZeroGPU for general text generation. Its hybrid sparse and linear attention keeps…
>
> [Open](https://docs.zerogpu.ai/api-reference/models/glm-5-3-flash.md)

> [card]
>
> ![qwen3-30b-a3b-fp8](https://models-favicon.zerogpu.ai/qwen3-30b-a3b-fp8/qwen3-30b-a3b-fp8.png)
>
> qwen3-30b-a3b-fp8
>
> 32,768 context window\$0.10 / 1M input\$0.45 / 1M output\$0.05 / 1M cached input
>
> Alibaba's Qwen3-30B-A3B is an open-weight Mixture-of-Experts model with 30.5B total parameters (3.3B active per token), served on ZeroGPU as an FP8 build for efficient inference. It thinks through a problem…
>
> [Open](https://docs.zerogpu.ai/api-reference/models/qwen3-30b-a3b-fp8.md)

> [card]
>
> ![qwen3.6-35b-a3b](https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg)
>
> qwen3.6-35b-a3b
>
> 35B params16,384 context window\$0.20 / 1M input\$1.50 / 1M output\$0.20 / 1M cached input
>
> Alibaba's Qwen3.6-35B-A3B is a fast, open-weight multimodal model for applications that need to understand text, images, and video, served on ZeroGPU as an FP8 build. It suits visual question answering, document and screenshot analysis…
>
> [Open](https://docs.zerogpu.ai/api-reference/models/qwen3-6-35b-a3b.md)

> [card]
>
> ![gpt-oss-120b](https://models-favicon.zerogpu.ai/gpt-oss-120b/gpt-oss-120b.png)
>
> gpt-oss-120b
>
> 131,072 context window\$0.15 / 1M input\$0.60 / 1M output\$0.03 / 1M cached input
>
> OpenAI's gpt-oss-120b is an open-weight Mixture-of-Experts model with 117B total parameters (5.1B active per token), served on ZeroGPU for general text generation. It reasons through a problem…
>
> [Open](https://docs.zerogpu.ai/api-reference/models/gpt-oss-120b.md)

> [card]
>
> ![llama-3.1-8b-instruct-fast](https://models-favicon.zerogpu.ai/llama-3.2-3b-instruct/meta.png)
>
> llama-3.1-8b-instruct-fast
>
> 131,072 max tokens\$0.15 / 1M input\$0.28 / 1M output\$0.025 / 1M cached input
>
> Meta's Llama 3.1 Instruct, tuned for fast, low-cost summarization at scale on the ZeroGPU edge network. Its 128K-token context window takes in entire documents, long transcripts, and full email or…
>
> [Open](https://docs.zerogpu.ai/api-reference/models/llama-3-1-8b-instruct-fast.md)

> [card]
>
> ![llama-3.2-3b-instruct-fast](https://models-favicon.zerogpu.ai/llama-3.2-3b-instruct/meta.png)
>
> llama-3.2-3b-instruct-fast
>
> 131,072 max tokens\$0.05 / 1M input\$0.30 / 1M output\$0.025 / 1M cached input
>
> Meta's Llama 3.2 3B Instruct, packaged by ZeroGPU for fast, low-cost summarization on edge GPUs. It shares the 131,072-token context window and the summarization pipeline of llama-3.1-8b-instruct-fast, and…
>
> [Open](https://docs.zerogpu.ai/api-reference/models/llama-3-2-3b-instruct-fast.md)

> [card]
>
> ![llama-guard-4-12b](https://models-favicon.zerogpu.ai/llama-3.2-3b-instruct/meta.png)
>
> llama-guard-4-12b
>
> Text GenerationSafety classificationText ModerationBrand Safety12B params163,840 context window\$0.18 / 1M input\$0.18 / 1M output
>
> Meta's Llama Guard 4 12B is a multimodal safety classification model for moderating text, images, and mixed text-image inputs. It evaluates both incoming prompts…
>
> [Open](https://docs.zerogpu.ai/api-reference/models/llama-guard-4-12b.md)

> [card]
>
> zlm-v1-tool-decisions
>
> 0.6B params24,576 context window\$0.05 / 1M input\$0.15 / 1M output
>
> ZeroGPU's tool-decision model turns user intent into action, pairing fast MCP tool selection with query and argument generation. It reads the conversation, selects the tools that fit, and generates the inputs needed to call them…
>
> [Open](https://docs.zerogpu.ai/api-reference/models/zlm-v1-tool-decisions.md)

> [card]
>
> unlimited-ocr
>
> 3B paramsDocument OCRPriced per page
>
> Baidu's Unlimited-OCR is a 3B-parameter document-parsing and OCR model, served on ZeroGPU as a bf16 build. It reads PDFs, scanned documents, and images and returns clean Markdown, with tables as HTML and formulas as LaTeX…
>
> [Open](https://docs.zerogpu.ai/api-reference/models/unlimited-ocr.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.