Does Ollama 0.34.1 break GGUF models I already have?

No. Ollama 0.34.1 changes how a GGUF model is created, not how an existing one runs — safetensor conversion and quantization now go through llama.cpp tooling. Existing GGUF models also keep support for the `typical_p` parameter, which is only removed for new models, so an installed library keeps working after the upgrade.

How much faster is the model list in Ollama 0.34.1?

The `/api/tags` endpoint, which Ollama clients call to list installed models, drops from a 3.1 second cold load to 294 milliseconds in version 0.34.1. That is roughly a tenfold improvement on the first call after start-up, which is the case users notice because it is the one that runs before anything is cached.

Why did Ollama change repeat token detection in 0.34.1?

Repeat token detection in Ollama now requires 100 tokens before it triggers. The previous threshold produced false positives, cutting off generations that legitimately repeated a short sequence — a table, a list, or structured output. Raising the threshold makes the guard more conservative, so it fires on real runaway loops rather than on normal repetition.

What else is in Ollama 0.34.1 for Apple Silicon users?

Beyond making MLX safetensors creation stable, Ollama 0.34.1 improves MLX memory handling on Apple Silicon and updates both the MLX and llama.cpp dependencies. Model capability reporting is also made consistent, so a client asking what a model supports gets the same answer regardless of which of the two backends is serving it.

Ollama · 2026-09-14 · major

Ollama 0.34.1 — MLX safetensors leave experimental, GGUF needs llama.cpp

Ollama 0.34.1 makes ollama create from MLX safetensors a stable feature. Building a GGUF model now requires llama.cpp tooling for safetensor conversion and quantization, and the /api/tags cold load falls from 3.1 seconds to 294 ms.

Ollama 0.34.1 promotes MLX safetensors model creation out of experimental and hands GGUF conversion to llama.cpp tooling.

Key specs

/api/tags cold load	3.1s to 294ms

Quick facts

Version	v0.34.1
Maker	Ollama
Released	2026-09-14
MLX safetensors	Stable in ollama create
GGUF creation	Now needs llama.cpp tooling
Repeat-token detection	Requires 100 tokens
Deprecated	typical_p for new models

What is it?

MLX safetensors support in `ollama create` is no longer experimental as of Ollama 0.34.1. The trade is on the other side: building a GGUF model now goes through llama.cpp tooling for safetensor conversion and quantization rather than through Ollama itself.

How does it work?

The two model formats now have separate owners inside Ollama — MLX handles safetensors on Apple Silicon, llama.cpp handles GGUF conversion and quantization. Version 0.34.1 also improves MLX memory handling on Apple Silicon, makes model capability reporting consistent, and raises the repeat-token detection threshold to 100 tokens so it stops triggering on short repeats.

Why does it matter?

Anyone who scripted `ollama create` against a GGUF build has a step to change, because that conversion now belongs to llama.cpp tooling. In exchange the model list gets much faster: a cold `/api/tags` load falls from 3.1 seconds to 294 milliseconds, and that is the call every Ollama client makes to find out which models are installed.

Who is it for?

people building custom local models

Frequently asked questions

Does Ollama 0.34.1 break GGUF models I already have?: No. Ollama 0.34.1 changes how a GGUF model is created, not how an existing one runs — safetensor conversion and quantization now go through llama.cpp tooling. Existing GGUF models also keep support for the `typical_p` parameter, which is only removed for new models, so an installed library keeps working after the upgrade.
How much faster is the model list in Ollama 0.34.1?: The `/api/tags` endpoint, which Ollama clients call to list installed models, drops from a 3.1 second cold load to 294 milliseconds in version 0.34.1. That is roughly a tenfold improvement on the first call after start-up, which is the case users notice because it is the one that runs before anything is cached.
Why did Ollama change repeat token detection in 0.34.1?: Repeat token detection in Ollama now requires 100 tokens before it triggers. The previous threshold produced false positives, cutting off generations that legitimately repeated a short sequence — a table, a list, or structured output. Raising the threshold makes the guard more conservative, so it fires on real runaway loops rather than on normal repetition.
What else is in Ollama 0.34.1 for Apple Silicon users?: Beyond making MLX safetensors creation stable, Ollama 0.34.1 improves MLX memory handling on Apple Silicon and updates both the MLX and llama.cpp dependencies. Model capability reporting is also made consistent, so a client asking what a model supports gets the same answer regardless of which of the two backends is serving it.

Try it

https://ollama.com/download