Skip to content
devhub.
ExploreCompareSaved
Sign in
DevHubBuilt for curious developers. Data from GitHub.
Back to repositories
Public repository / P-kaizoku

ollama

Get up and running with Kimi-K2.6, GLM-5.2, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.

View on GitHub
0Stars
0Forks
0Open issues
0Subscribers

Language distribution

Share of source bytes reported by GitHub

Go71%
C20%
TypeScript5%
C++1%
Objective-C1%
CMake1%
PowerShell<1%
Shell<1%
Dockerfile<1%
Inno Setup<1%
Go Template<1%
HTML<1%
AppleScript<1%
CSS<1%
JavaScript<1%

Recent activity

Latest commits on the default branch

app: preserve Codex configuration across host changes (#18247)ParthSareen · Sep 8, 2026openai: use tsc_ prefix for tool search call IDs (#18296)jmorganca · Sep 8, 2026app: harden Codex desktop proxy handling (#18244)ParthSareen · Sep 5, 2026openai: preserve images through response compaction (#18245)ParthSareen · Sep 5, 2026app: route Codex auto review through selected model (#18246)ParthSareen · Sep 5, 2026ci: bound and cache the macOS MLX payload build (#18240)dhiltgen · Sep 4, 2026app: add Ollama to ChatGPT Desktop (#18236)hoyyeva · Sep 4, 2026

Project details

Created Sep 2026 Updated Sep 2026 MIT License Project website

Contributors

8
mxyng's avatarmxyng1294 commitsdhiltgen's avatardhiltgen972 commitsjmorganca's avatarjmorganca916 commitsBruceMacD's avatarBruceMacD364 commitsjessegross's avatarjessegross293 commitsParthSareen's avatarParthSareen228 commitspdevine's avatar
openai: add Codex compaction support (#18224)ParthSareen · Sep 4, 2026
openai: add client tool search support (#18223)ParthSareen · Sep 4, 2026
mlxrunner: compile structured output as xgrammar structural tagsjessegross · Sep 3, 2026
pdevine
215 commits
technovangelist's avatartechnovangelist127 commits