Moshi 7B

Overview

Kyutai's full-duplex speech model — 7.6B parameters with sub-second latency (~200ms) and two voices, Moshiko and Moshika. A speech architecture, not a text LLM.

When to pick this model

• You're building real-time voice interfaces and need full-duplex behavior
• You need low-latency speech-to-speech without separate TTS and STT
• You're researching speech architectures rather than text LLMs
• You can run inference directly in PyTorch or via Kyutai's stack

VRAM requirements by quantization

Quantization	VRAM required
Q4_K_M (recommended)	5 GB
Q5_K_M	6 GB
Q8_0	9 GB
FP16 (no quantization)	15 GB

VRAM figures include model weights plus a typical 8k KV cache and ~600 MB runtime overhead (Ollama / llama.cpp baseline). Add headroom for higher context lengths.

In practice, Moshi 7B fits an 8 GB consumer card at Q4_K_M (5 GB). Stepping up to Q8_0 nearly doubles the footprint to 9 GB, and unquantized FP16 weights take 15 GB — plan your GPU around the Q4 or Q5 figure unless you specifically need the higher fidelity.

Without a GPU, Moshi 7B needs roughly 10 GB of system RAM to run on CPU via llama.cpp or Ollama — workable for background jobs, but far slower than GPU inference. Throughput estimates from our compatibility engine: around 8 tokens/sec on entry-level GPUs, on the order of 25 tokens/sec on a mid-range card, and up to 60 tokens/sec on high-end hardware — assuming the chosen quantization fully fits in VRAM.

What hardware do you need

The table below matches Moshi 7B to common GPU memory tiers, using the highest-fidelity quantization that fully fits each card class. Spilling layers to system RAM works but costs most of the speed, so size your card to the quantization you actually want to run.

GPU memory	Example cards	Best fit for Moshi 7B
8 GB	RTX 5070 Laptop, RTX 5060, RTX 5060 Ti 8GB	Q5_K_M (6 GB used)
12 GB	RTX 5070, RTX 5070 Ti Laptop, RTX 4080 Laptop	Q8_0 (9 GB used)
16 GB	RTX 5080, RTX 4080 Super, Radeon RX 9070 XT	FP16 (15 GB used)
24 GB	RTX 4090, Radeon RX 7900 XTX, RTX 5090 Laptop	FP16 (15 GB used)
32 GB	RTX 5090	FP16 (15 GB used)

Which GPU should you buy to run Moshi 7B?

To run Moshi 7B locally at Q4, you need ~5 GB of VRAM. The best value for this is a RTX 5060 (8 GB VRAM).

Strengths

• First open full-duplex speech model
• Sub-second latency (~200ms in practice)
• Mimi codec at 12.5 Hz / 1.1 kbps on 24 kHz audio
• From Kyutai, a respected French AI lab

Limitations

• Not a text LLM — different use case entirely
• Architecture not supported by Ollama
• CC-BY 4.0 license — attribution required

Typical workloads

In our catalog grid, Moshi 7B is filed under Voice Conversation, Real-Time — the use cases where its size/quality trade-off makes the most sense. Its tags translate to concrete workloads: audio understanding; French-language output where quality matters.

Note the 4k-token context window — fine for short interactions, limiting for long documents or big retrieval contexts. It ships under the CC-BY 4.0 license — commercial use is generally possible but read the specific terms before embedding it in a product.

Architecture & training

Architecture: Full-duplex speech-text · Depth Transformer (codebook) + 7B Temporal Transformer

Training: Mimi codec at 12.5 Hz / 1.1 kbps on 24 kHz audio. ~200ms practical latency.

Quick start

# GitHub : kyutai-labs/moshi — voix Moshiko (H) / Moshika (F)

Or use the open-source MCP server to query this model from Claude Desktop, Cursor, or any MCP-compatible client.

Similar models worth comparing

Frequently asked questions

How much VRAM does Moshi 7B need?

At the recommended Q4_K_M quantization, Moshi 7B needs about 5 GB of VRAM. Q8_0 takes 9 GB, and unquantized FP16 weights take 15 GB.

Can Moshi 7B run without a GPU?

Yes — with roughly 10 GB of system RAM it runs CPU-only through llama.cpp or Ollama. Expect a fraction of GPU speed, which is fine for background or batch jobs but slow for interactive chat.

What context window does Moshi 7B support?

Moshi 7B supports a 4k-token context window (4,096 tokens).

Can I use Moshi 7B commercially?

Moshi 7B ships under the CC-BY 4.0 license. Commercial use is generally permitted subject to its terms — review the license text before shipping a product.

How fast is Moshi 7B on consumer hardware?

Our compatibility engine estimates on the order of 25 tokens/sec on a mid-range GPU and up to 60 tokens/sec on high-end cards, assuming the quantization fully fits in VRAM.

Which quantization of Moshi 7B should I download first?

Start with Q4_K_M (5 GB) — the standard size/quality sweet spot. Step up to Q5_K_M or Q8_0 only if you have VRAM headroom. It already fits an 8 GB card at Q5_K_M.