🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10227
📅 Released: 2026-08-02 09:43 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

chat : add qwen3 specialized parser (#26252)

Add tagged thinking tool parser

chat : refactor and add permute helper

cont : add support for <tool_call> omission

cont : update tool delimiters

cont : add comment for qwen3-coder

cont : fix trigger pattern for <function

---------

Co-authored-by: Bart de Boer <bart.deboer@gmail.com>

</details>

Website:
- <htt...

🔗 View on GitHub
📥 Download Release

🚀 New Release Alert!

📦 thefeed
🏷 Category: Community Suggested
🔖 Version: v0.38.0
📅 Released: 2026-08-02 11:01 UTC
📊 Assets: 27 files

📝 What's New:
## Install


Android and iOS install...

🔗 View on GitHub
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10228
📅 Released: 2026-08-02 13:28 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

DeepseekV4 MTP + DSpark (#25784)

</details>

Website:
- <https://llama.app>

macOS/iOS:
- macOS Apple Silicon (arm64)
- macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED
- macOS Intel (x64)
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10229
📅 Released: 2026-08-02 14:31 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

opencl: bugfix increment refcount in ggmlbackendopenclinit() (#26162)

Incrementing ref_count at the beginning is important later
in the free() method of the ggml_backend_opencl_context at program end.
If we do not increment the ref_count, the result would be -1 here,
and consequently, the profiling data would not be flushed and written.
( #ifdef GGMLOPENCLPROFILI...

🔗 View on GitHub
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10231
📅 Released: 2026-08-02 18:14 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

common: support the DSpark sidecar resolution (#26458)

The dspark- files resolve like the other speculative sidecars: the
-hfd tag applies to them, a requested sidecar resolves without a full
model at the tag, and an explicit -md selection disables the discovery.
When no type is requested, dspark outranks dflash in the auto-selection
since its sidecar carries the extra Markov h...

🔗 View on GitHub
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10232
📅 Released: 2026-08-02 18:57 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

metal: implement DeepSeek V4 hyper-connections (#26459)

- Implement GGMLOPDSV4HCCOMB, GGMLOPDSV4HCPRE, and
GGMLOPDSV4HCPOST with SIMDgroup register and shuffle optimized kernels.
- Add Metal dispatch and support plumbing and test the production Sinkhorn
iteration count and embedding width.

Assisted-by: Codex

Co-authored-by: Thiago Padilha <thiago@padilha.cc>

...

🔗 View on GitHub
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10234
📅 Released: 2026-08-02 20:17 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

metal : add F16 support for bin ops (#26465)

</details>

Website:
- <https://llama.app>

macOS/iOS:
- macOS Apple Silicon (arm64)
- macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED
- macOS Intel (x64)
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10235
📅 Released: 2026-08-02 21:02 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

metal : add SILUBACK (#25982)

* feat(siluback): implemented siluback op for f32

* fix(siluback): removed redundant asserts in ggml-metal-ops.cpp function ggmlmetalopsiluback.

</details>

Website:
- <https://llama.app>

macOS/iOS:
- macOS Apple Silicon (arm64)
...

🔗 View on GitHub
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10237
📅 Released: 2026-08-03 07:50 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

llama : MTP support for DeepSeek V3.2 (#26457)

llama : MTP support for DeepSeek V3.2

model : no need to include MTP layers during DeepSeek V3.2 model type discovery

---------

Co-authored-by: Stanisław Szymczyk <sszymczy@gmail.com>

</details>

Website:
- <https://llama.app>

macOS/iOS:
- macOS Apple Silicon (arm64)
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10238
📅 Released: 2026-08-03 08:56 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

model: MTP support for Qwen3-Next (#25589)

mtp for qwen3nex

fix for python type-check

Fix to compute num_mtp from directly mtp layer

define optnummtplayers in QwenMtpMixin and fix some comments

Fix for python type check

Update gguf-py/gguf/constants.py

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co>

rebase and add load_mtp flags

U...

🔗 View on GitHub
📥 Download Release

🚀 New Release Alert!

📦 OpenFrontIO
🏷 Category: Community Suggested
🔖 Version: v0.33.0
📅 Released: 2026-08-03 17:19 UTC
📊 Assets: 0 files

📝 What's New:
# v0.33.0

Big one. New gamemode, ranked 2v2, 22 new maps, and mappers can finally block off terrain.

## Overview

Ranked 2v2
Ranked isn't 1v1-only anymore. Separate ladder, and it'll try to put you with your clanmates. Lobbies that don't fill or don't fully spawn now cancel instead of wasting your time.

Login with Google
Sign in with Google, link it to the account you alrea...

🔗 View on GitHub
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10243
📅 Released: 2026-08-03 20:10 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

llama : allocate indexer cache only in "full" indexer layers (#26474)

Co-authored-by: Stanisław Szymczyk <sszymczy@gmail.com>

</details>

Website:
- <https://llama.app>

macOS/iOS:
- macOS Apple Silicon (arm64)
- macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED...

🔗 View on GitHub
📥 Download Release

🚀 New Release Alert!

📦 OpenFrontIO
🏷 Category: Community Suggested
🔖 Version: v0.33.1
📅 Released: 2026-08-03 21:39 UTC
📊 Assets: 0 files

📝 What's New:
# v0.33.1

bugfix: tribes not showing up in leaderboard
bugfix: replays showing desync error

# v0.33.0

Big one. New gamemode, ranked 2v2, 22 new maps, and mappers can finally block off terrain.

## Overview

Ranked 2v2
Ranked isn't 1v1-only anymore. Separate ladder, and it'll try to put you with your clanmates. Lobbies that don't fill or don't fully spawn now cancel instea...

🔗 View on GitHub
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10245
📅 Released: 2026-08-03 22:57 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

graph : fix unused input tensors in minimax m3 graph (#26519)

</details>

Website:
- <https://llama.app>

macOS/iOS:
- macOS Apple Silicon (arm64)
- macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED
- macOS Intel...

🔗 [View on GitHub
📥 Download Release

🚀 New Release Alert!

📦 claude-code
🏷 Category: Community Suggested
🔖 Version: v2.1.221
📅 Released: 2026-08-04 00:14 UTC
📊 Assets: 10 files

📝 What's New:
## What's changed

- VSCode Added Focus view: a chat-menu toggle that hides tool activity behind an expandable per-turn summary with a live running-tool indicator, toggled with Ctrl+Alt+F or the "Claude Code: Toggle Focus view" command
- Added mode: "mask" for sandbox credential files on Linux and WSL — sandboxed commands read a sentinel copy (the whole file, or just the spans captured by...

🔗 View on GitHub
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10248
📅 Released: 2026-08-04 01:01 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

vocab : validate default special token ids (#26506)

</details>

Website:
- <https://llama.app>

macOS/iOS:
- macOS Apple Silicon (arm64)
- macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED
- macOS Intel (x64)
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10251
📅 Released: 2026-08-04 03:06 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

model : support MTP in GLM-4.7-Flash (#24868)

</details>

Website:
- <https://llama.app>

macOS/iOS:
- macOS Apple Silicon (arm64)
- macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED
- macOS Intel (x64)
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10253
📅 Released: 2026-08-04 04:22 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

vendor : update cpp-httplib to 0.52.0 (#26485)

</details>

Website:
- <https://llama.app>

macOS/iOS:
- macOS Apple Silicon (arm64)
- macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED
- macOS Intel (x64)
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10258
📅 Released: 2026-08-04 09:22 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

llama : move nvocab from llamasamplerdata to penaltysampler (#26520)

This matches how it is done for logitbias and mirostat samplers, see
https://github.com/ggml-org/llama.cpp/pull/25262#discussionr3703951151

</details>

Website:
- <https://llama.app>

macOS/iOS:
- macOS Apple Silicon (arm64)
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10259
📅 Released: 2026-08-04 12:18 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

model : allow reshape of tensors during load (#26531)

</details>

Website:
- <https://llama.app>

macOS/iOS:
- macOS Apple Silicon (arm64)
- macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED
- macOS Intel (x64)
📥 Download Release

🚀 New Release Alert!

📦 llama.cpp
🏷 Category: AI
🔖 Version: b10261
📅 Released: 2026-08-04 13:09 UTC
📊 Assets: 25 files

📝 What's New:
<details open>

vocab : validate plamo2 byte tokens (#26511)

validate plamo2 byte tokens

--typo

</details>

Website:
- <https://llama.app>

macOS/iOS:
- macOS Apple Silicon (arm64)
- macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED
📥 Download Release