# RWKV7-G1k-13.3B-20260930

> This is an official BlinkDL release of RWKV-7 Goose in Hugging Face Transformers format. RWKV-7 is an attention-free recurrent architecture with a constant-size recurrent state and constant inference work per generated token. Training remains parallelizable.

- **Organisation:** RWKV
- **Category:** Chat & Text Generation
- **Parameters:** 13.3B
- **Licence:** Apache 2.0 (Commercial use OK)
- **Released:** 2026-09-30
- **Hugging Face:** 438 downloads (30 days), 8 likes

## Estimated memory

| Quant | ≈ GB |
|---|---|
| Q2_K | 12.39 |
| Q3_K_M | 14.47 |
| Q4_K_M | 15.99 |
| Q5_K_M | 17.35 |
| Q6_K | 18.81 |
| Q8_0 | 21.91 |
| F16 | 34.08 |

## Runs with

- [llama.cpp](https://selfhost.directory/project/llamacpp)
- [Transformers](https://github.com/huggingface/transformers)

## About

This is an official BlinkDL release of RWKV-7 Goose in Hugging Face Transformers format. RWKV-7 is an attention-free recurrent architecture with a constant-size recurrent state and constant inference work per generated token. Training remains parallelizable.

This checkpoint is a base model pretrained with web, code, synthetic, instruction, chat, and reasoning data. It is suitable for evaluation, post-training, and fine-tuning; the included chat template is a prompt interface, not a claim that the checkpoint is a safety-aligned assistant.

The Transformers integration, conversion, release packaging, linear-time RWKV tokenizer, and optional TileLang inference implementation are distributed with this release.

• Constant recurrent state: memory does not grow like an attention KV cache.
• Bundled Transformers integration: auditable remote configuration and modeling
• modules provide generation, recurrent cache continuation, training, and LoRA
• workflows on Transformers 5.15+.
• Exact linear-time tokenizer: the bundled RWKV trie reads the self-contained
• tokenizer.json generated from the canonical RWKV World byte vocabulary.
• Chat-ready: chat_template.jinja supports system, multi-turn, thinking, and
• strict model-generated tool-call prompts.
• Optional optimized runtime: the isolated inference/ bundle
• provides PyTorch fallback and TileLang acceleration without changing the
• standard model root.

## Alternatives

- [Qwen/Qwen3-14B](https://selfhost.directory/models/qwen/qwen3-14b) (14.8B, Apache 2.0, ≈10.51 GB at Q4): Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models.
- [ornith-ai/Ornith-1.5-9B](https://selfhost.directory/models/ornith-ai/ornith-1.5-9b) (9.7B, MIT, ≈7.22 GB at Q4): Chirp Chirp! 🐦 We are introducing Ornith-1.5, a major step toward building foundation models through end-to-end self-improvement.
- [ornith-ai/Ornith-1.0-9B](https://selfhost.directory/models/ornith-ai/ornith-1.0-9b) (9B, MIT, ≈6.84 GB at Q4): Aloha! 🌺 Today, we are releasing Ornith-1.0, a self-improving family of open-source models for agentic coding.
- [openai/gpt-oss-20b](https://selfhost.directory/models/openai/gpt-oss-20b) (20.9B (4.2B active), Apache 2.0, ≈13.27 GB at Q4): Welcome to the gpt-oss series, OpenAI’s open-weight models designed for powerful reasoning, agentic tasks, and versatile developer use cases.
- [speakleash/Bielik-11B-v3.0-Instruct](https://selfhost.directory/models/speakleash/bielik-11b-v3.0-instruct) (11.2B, Apache 2.0, ≈8.33 GB at Q4): Bielik-11B-v3.0-Instruct is a generative text model featuring 11 billion parameters. It is an instruct fine-tuned version of the Bielik-11B-v3-Base-20250730.
- [nvidia/Nemotron-Labs-Diffusion-14B](https://selfhost.directory/models/nvidia/nemotron-labs-diffusion-14b) (13.5B, nvidia-nemotron-open-model-license, ≈9.76 GB at Q4): Nemotron-Labs-Diffusion is a tri-mode language model that supports both AR decoding and diffusion-based parallel decoding by simply switching the attention pattern of the same model during inference.

More: https://selfhost.directory/models/rwkv/rwkv7-g1k-13.3b-20260930/alternatives

## Runs well on

- [Tesla P40](https://selfhost.directory/models/rwkv/rwkv7-g1k-13.3b-20260930/on/tesla-p40): Q8_0, est. 14 tok/s
- [Radeon RX 7900 XTX](https://selfhost.directory/models/rwkv/rwkv7-g1k-13.3b-20260930/on/rx-7900-xtx): Q8_0, est. 37 tok/s
- [Instinct MI50 32GB](https://selfhost.directory/models/rwkv/rwkv7-g1k-13.3b-20260930/on/instinct-mi50): Q8_0, est. 40 tok/s
- [GeForce RTX 5090](https://selfhost.directory/models/rwkv/rwkv7-g1k-13.3b-20260930/on/rtx-5090): Q8_0, est. 70 tok/s
- [Apple M4 Pro (48 GB)](https://selfhost.directory/models/rwkv/rwkv7-g1k-13.3b-20260930/on/apple-m4-pro-48gb): F16, est. 5.7 tok/s
- [RTX A6000](https://selfhost.directory/models/rwkv/rwkv7-g1k-13.3b-20260930/on/rtx-a6000): F16, est. 16 tok/s

## Similar size

- [Qwen/Qwen3-14B](https://selfhost.directory/models/qwen/qwen3-14b) (14.8B)
- [xtools-at/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic](https://selfhost.directory/models/xtools-at/deepseek-v4-pro-qwen3.5-9b-uncensored-heretic) (9.4B)
- [microsoft/phi-4](https://selfhost.directory/models/microsoft/phi-4) (14.7B)
- [ornith-ai/Ornith-1.5-9B](https://selfhost.directory/models/ornith-ai/ornith-1.5-9b) (9.7B)
- [Qwen/Qwen2.5-14B-Instruct](https://selfhost.directory/models/qwen/qwen2.5-14b-instruct) (14.8B)
- [openbmb/MiniCPM-SALA](https://selfhost.directory/models/openbmb/minicpm-sala) (9.5B)

## FAQ

**How much VRAM does RWKV7-G1k-13.3B-20260930 need?**
About 16 GB at Q4_K_M with an 8K context. Q8_0 needs about 22 GB and full precision about 34 GB. Longer contexts need more memory for the KV cache.

**Can RWKV7-G1k-13.3B-20260930 run on the Tesla P40?**
Yes. It fits at Q8_0 (about 22 GB), at an estimated 14 tokens/s. It fits comfortably on the Radeon RX 7900 XTX, Instinct MI50 32GB, RTX 5090.

**Can I use RWKV7-G1k-13.3B-20260930 commercially?**
Yes. It is released under Apache 2.0. Permissive: use, modify and sell freely; keep the notice and licence text.

**How do I run RWKV7-G1k-13.3B-20260930 locally?**
It runs with Transformers. Each has copy-paste examples in its docs.

## Links

- huggingface: https://huggingface.co/RWKV/RWKV7-G1k-13.3B-20260930
- paper: https://arxiv.org/abs/2503.14456

---
Source: https://selfhost.directory/models/rwkv/rwkv7-g1k-13.3b-20260930