Every commit scored for change-risk against this repo's own history, so 'elevated' means elevated here rather than on some global curve.
Needs review
1,610 commits sit in this repo's top risk tercile, which is 34% of the 4,784scored. The cut is drawn against this codebase's own history rather than a global curve, so a quiet repo still fills its top band, and here it starts at 5.7 out of 10. What pushes a commit up is size and spread together: a large change confined to one area scores below a smaller one scattered across a dozen files.
Commit categories over time, read off the subject line. Fixes carry the accent because that is the series this chart exists to show.
Consistently other-driven across its history.
Ranked by change-risk, highest first. Priority is a tercile of this repo's own distribution, so a quiet repo still fills its top band.
| # | Commit | Author | When | Lines | Risk | Top driver |
|---|---|---|---|---|---|---|
| 1 | 0cf7794bggml update to b7108 (#12992) | Daniel Hiltgen | 8mo ago | +33.0K -23.7K | 100%Elevated | more lines added than baseline |
| 2 | d3b4b997app: add code for macOS and Windows apps under 'app' (#12933) | Daniel Hiltgen | 9mo ago | +103.0K -1.5K | 100%Elevated | more lines added than baseline |
| 3 | 1a19df1fupdate vendored llama.cpp and ggml (#11823) | Michael Yang | 11mo ago | +156.8K -48.3K | 100%Elevated | more lines added than baseline |
| 4 | 4f473e22int: add performance integration tests (#11173) | Daniel Hiltgen | 1y ago | +124.8K -31 | 100%Elevated | more lines added than baseline |
| 5 | 943464ccllama: update to commit 71e90e88 (#10192) | Jeffrey Morgan | 1y ago | +43.0K -33.9K | 100%Elevated | more lines added than baseline |
| 6 | 58245413next ollama runner (#7913) | Michael Yang | 1y ago | +475.4K -493 | 100%Elevated | more lines added than baseline |
| 7 | 527cc978llama: update vendored code to commit 40c6d79f (#7875) | Jeffrey Morgan | 1y ago | +62.5K -45.8K | 100%Elevated | more lines added than baseline |
| 8 | 96efd905Re-introduce the `llama` package (#5034) | Jeffrey Morgan | 1y ago | +166.1K -166 | 100%Elevated | more lines added than baseline |
| 9 | 9ac6440dImport server.cpp as of b2356 | Daniel Hiltgen | 2y ago | +43.3K -0 | 100%Elevated | more lines added than baseline |
| 10 | 442dec1cvendor llama.cpp | Michael Yang | 3y ago | +35.2K -62 | 100%Elevated | more lines added than baseline |
| 11 | 82f905cdcmd: agent UI (#17017) | Parth Sareen | 4w ago | +14.8K -192 | 100%Elevated | more lines added than baseline |
| 12 | 964ea42cmlx: x/create rewrite (#16919) | Patrick Devine | 1mo ago | +3.7K -5.5K | 100%Elevated | more lines added than baseline |
| 13 | a2b3a5e9agent: harness core (#16963) | Parth Sareen | 1mo ago | +6.6K -3.9K | 100%Elevated | more lines added than baseline |
| 14 | 9db4bdbarunner: Remove CGO engines, use llama-server exclusively for GGML models (#16031) | Daniel Hiltgen | 2mo ago | +28.4K -430.0K | 100%Elevated | more lines added than baseline |
| 15 | 87288cedNew models (#15861) | Daniel Hiltgen | 3mo ago | +11.3K -633 | 100%Elevated | more lines added than baseline |
| 16 | af5f7c0acmd: refactor tui and launch (#14609) | Parth Sareen | 4mo ago | +5.2K -3.3K | 100%Elevated | more lines added than baseline |
| 17 | 5f5ef201anthropic: enable websearch (#14246) | Parth Sareen | 5mo ago | +4.3K -31 | 100%Elevated | more lines added than baseline |
| 18 | 44bdd9a2Add MLX runner with GLM4-MoE-Lite model support (#14185) | Patrick Devine | 5mo ago | +14.9K -9 | 100%Elevated | more lines added than baseline |
| 19 | ef00199fUpdate vendor ggml code to a5bb8ba4 (#13832) | Jeffrey Morgan | 6mo ago | +21.2K -5.0K | 100%Elevated | more lines added than baseline |
| 20 | 199c41e1cmd: `ollama config` command to help configure integrations to use Ollama (#13712) | Parth Sareen | 6mo ago | +4.5K -0 | 100%Elevated | more lines added than baseline |
| 21 | 33ee7168Add experimental MLX backend and engine with imagegen support (#13648) | Daniel Hiltgen | 7mo ago | +27.5K -88 | 100%Elevated | more lines added than baseline |
| 22 | 3a9e8e9fvulkan: temporary cary of vulkan fixes (#12971) | Daniel Hiltgen | 8mo ago | +5.8K -642 | 100%Elevated | more lines added than baseline |
| 23 | 3d99d977docs: add docs for docs.ollama.com (#12805) | Parth Sareen | 9mo ago | +5.0K -2.2K | 100%Elevated | more lines added than baseline |
| 24 | 934dd9e1docs: add reference to docs.ollama.com (#12800) | Parth Sareen | 9mo ago | +6.7K -0 | 100%Elevated | more lines added than baseline |
| 25 | 2aba569aVulkan based on #9650 (#11835) | Thomas Stocker | 9mo ago | +29.4K -15 | 100%Elevated | more lines added than baseline |
| 26 | c68f367eUpdate GGML to b6646 (#12245) | Daniel Hiltgen | 10mo ago | +30.7K -20.7K | 100%Elevated | more lines added than baseline |
| 27 | 0cefd46fllama: update to commit de4c07f93 (#10655) | Jeffrey Morgan | 1y ago | +8.0K -4.2K | 100%Elevated | more lines added than baseline |
| 28 | d7d7e996llama: update llama.cpp vendor code to commit d7cfe1ff (#9356) | Jeffrey Morgan | 1y ago | +18.2K -11.0K | 100%Elevated | more lines added than baseline |
| 29 | 1deafd82llama: update vendored code to commit 46e3556 (#8308) | Jeffrey Morgan | 1y ago | +23.7K -20.6K | 100%Elevated | more lines added than baseline |
| 30 | f2890a44IBM granite/granitemoe architecture support (#6760) | Gabe Goodhart | 1y ago | +14.2K -10.8K | 100%Elevated | more lines added than baseline |
| 31 | b1c88eb9add `privategpt` example | Jeffrey Morgan | 3y ago | +5.8K -0 | 100%Elevated | more lines added than baseline |
| 32 | 32c6d8e6Use single server | Jeffrey Morgan | 3y ago | +11.9K -10.0K | 100%Elevated | more lines added than baseline |
| 33 | 2f4aa42dExploratory client | Jeffrey Morgan | 3y ago | +6.6K -89 | 100%Elevated | more lines added than baseline |
| 34 | a51df815test: revamp integration test entrpoints (#16560) | Daniel Hiltgen | 2w ago | +1.6K -1.3K | 99%Elevated | more lines added than baseline |
| 35 | d0062206mlxrunner: drive MTP speculation through cache snapshots | Jesse Gross | 2mo ago | +2.1K -550 | 99%Elevated | more lines added than baseline |
| 36 | 2bbe2405mlxrunner: decouple models from attention cache storage layout | Jesse Gross | 3mo ago | +2.7K -331 | 99%Elevated | more lines added than baseline |
| 37 | 43f90deflaunch: add hermes (#15569) | Parth Sareen | 3mo ago | +3.0K -110 | 99%Elevated | more lines added than baseline |
| 38 | 2cba7756Gemma4 on MLX (#15244) | Daniel Hiltgen | 3mo ago | +2.7K -0 | 99%Elevated | more lines added than baseline |
| 39 | 96b202d3Add support for gemma4 (#15214) | Daniel Hiltgen | 4mo ago | +7.2K -51 | 99%Elevated | more lines added than baseline |
| 40 | 96e36c0dmlxrunner: share KV cache across conversations with common prefixes | Jesse Gross | 4mo ago | +2.7K -171 | 99%Elevated | more lines added than baseline |
| 41 | 4b2ac1f3model: improvements to LFM architectures (#14368) | Jeffrey Morgan | 5mo ago | +3.3K -1.3K | 99%Elevated | more lines added than baseline |
| 42 | f08427c1cmd: TUI UX improvements (#14198) | Parth Sareen | 5mo ago | +2.1K -1.9K | 99%Elevated | more lines added than baseline |
| 43 | b1fccabbRevert "Update vendored llama.cpp to b7847" (#14061) | Jeffrey Morgan | 6mo ago | +5.0K -21.2K | 99%Elevated | more lines added than baseline |
| 44 | d8cc798cglm 4.7 flash support on experimental engine (#13838) | Patrick Devine | 6mo ago | +3.9K -2.4K | 99%Elevated | more lines added than baseline |
| 45 | a887406cx/imagegen: add preliminary support for FLUX.2-klein model (#13772) | Jeffrey Morgan | 6mo ago | +3.3K -380 | 99%Elevated | more lines added than baseline |
| 46 | 12719b6eMLX - dynamic loading of mlx-c (#13735) | Daniel Hiltgen | 6mo ago | +9.0K -228 | 99%Elevated | more lines added than baseline |
| 47 | a077d996Fix `create` and `show` commands for experimental models (#13741) | Patrick Devine | 6mo ago | +2.6K -266 | 99%Elevated | more lines added than baseline |
| 48 | 25849400Add z-image image generation prototype (#13659) | Jeffrey Morgan | 7mo ago | +6.4K -264 | 99%Elevated | more lines added than baseline |
| 49 | 2b2cda7aapi: implement anthropic api (#13600) | Parth Sareen | 7mo ago | +3.0K -3 | 99%Elevated | more lines added than baseline |
| 50 | 49a9c9baGGML update to ec98e2002 (#13451) | Daniel Hiltgen | 7mo ago | +8.2K -6.7K | 99%Elevated | more lines added than baseline |
Two views of the same model: where the cuts fall, and what commit shape lands you above them.
Every scored commit, binned on the raw 0 to 10 score rather than the percentile. Percentile ranks are uniform by construction, so that axis has no shape to draw. The dashed lines are the tercile cuts behind each row's priority pill.
The 200 most recent commits, on their own recency sample rather than the feed above: that defaults to risk-sorted, so reusing it would plot only the top tercile and call it the spread. Big and scattered is what the model penalises. Click a dot to open it.
ollama/ollama has 5,582 commits in its history from 614 contributors, the first of them Jun 22, 2023. In the last 90 days 482 files were touched, 1,107 times in total, most often integrations_test.go. Every commit is scored for change risk from its size, spread and the history of the files it touches.
