Pull Request Explorer
| Title | Author | Size | AI | Cycle Time | Review | Merged |
|---|---|---|---|---|---|---|
| cli: add decision model support to `ollama run` | jmorganca | XL | AI | 23.3h | 21.8h | Oct 13, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| Add decision model integration tests and benchmarks | dhiltgen | XL | No | 278.2h | 264.4h | Oct 13, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| gemma4: use MLX SDPA for CUDA prefill with wide head dims | dhiltgen | S | No | 187.6h | 176.4h | Oct 13, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| ci: harden download steps with retries | dhiltgen | M | No | 115.2h | 104.0h | Oct 13, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| mlx: preserve generation panics during cleanup | dhiltgen | M | No | 114.3h | 103.0h | Oct 13, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| Add Kev decision scoring to MLX | dhiltgen | XL | No | 0.8h | 0.1h | Oct 13, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| server: fix multimodal embedding load with small num_ctx | jmorganca | S | No | 0.2h | 0.1h | Oct 13, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| docs: update SECURITY.md | anra8571 | XS | No | 1.4h | 0.2h | Oct 12, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| Revert "docs: restructuring" | jmorganca | XL | No | 0.0h | - | Oct 12, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| docs: restructuring | anra8571 | XL | No | 138.4h | 0.7h | Oct 12, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| MLX, MLX-C: version bump | dhiltgen | M | No | 1.4h | 1.4h | Oct 12, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| launch: include installed models in Codex /model picker | ParthSareen | M | No | 90.7h | 90.7h | Oct 12, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| mlx: compact restored recurrent state after speculative rollback | dhiltgen | L | No | 0.4h | - | Oct 12, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| server: reuse verified manifests and loaded models in inference routes | dhiltgen | XL | No | 18.2h | 1.9h | Oct 12, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| llama: update llama.cpp to b11521 for decision models and EmbeddingGemma 2 | jmorganca | XL | No | 48.3h | 0.1h | Oct 12, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| server: skip local model compatibility migration | jmorganca | S | No | 0.5h | - | Oct 09, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| launch: suppress Claude Code connector warning for Ollama | hoyyeva | S | AI | 21.5h | 2.5h | Oct 09, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| launch: avoid inherited service-tier warnings in Codex | hoyyeva | XS | No | 2.6h | 2.4h | Oct 08, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| Set CLAUDE_CODE_MAX_CONTEXT_TOKENS to model's context length | jberg5 | M | No | 18.6h | 0.0h | Oct 08, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| server: hide duplicate and downgrade guards from list | dhiltgen | S | No | 0.3h | 0.1h | Oct 08, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| README: add oxi to community integrations | maziluiosif | XS | No | 145.8h | - | Oct 07, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| mlx: drop carried metal residency patch now that it is upstream | dhiltgen | M | No | 1.2h | 1.2h | Oct 07, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| docs: fix 6 dead links in README community integrations list | aniketkrs | S | No | 39.9h | - | Oct 07, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| manifest: avoid symlinks on Windows | dhiltgen | M | No | 0.4h | 0.4h | Oct 07, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| cmd: remove account step from CLI onboarding | hoyyeva | L | No | 21.5h | 3.1h | Oct 07, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| llama: fix clef head reads past 2GiB on windows | Gigrise | S | No | 79.8h | 79.8h | Oct 07, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| server: proxy cloud usage and balance APIs | drifkin | XL | No | 0.5h | 0.5h | Oct 07, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| routes: fix embed 413 vs 400 routing | pdevine | S | No | 0.2h | 0.2h | Oct 06, 2026 |
SummaryHealth AssessmentAI DetailsTech Stack |
||||||
| model: add multimodal embeddings | pdevine | XL | No | 0.2h | 0.0h | Oct 06, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| mlx: fix patch for recent mlx update | dhiltgen | XS | No | 0.3h | 0.3h | Oct 06, 2026 |
SummaryHealth AssessmentAI DetailsTech Stack |
||||||
| MLX: version bump | dhiltgen | XS | No | 127.2h | 126.7h | Oct 06, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| llama.cpp: version update | dhiltgen | L | No | 75.7h | 75.3h | Oct 06, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| Reduce model lookup and MLX decision request overhead | dhiltgen | L | No | 4.1h | 3.5h | Oct 06, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| mlx: mitigate high latency after GPU idle | dhiltgen | M | No | 3.0h | 2.6h | Oct 05, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| openai: keep tool message content parts in one message | dongluochen | M | No | 121.3h | 25.5h | Oct 05, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| pull: allow RCs to pull matching min_version | dhiltgen | S | No | 0.6h | 0.0h | Oct 04, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| mlx: match publisher tokenizer semantics | dhiltgen | XL | No | 14.2h | 3.5h | Oct 04, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| mlx: Decision model improvements | dhiltgen | L | No | 4.4h | 4.4h | Oct 04, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| mlx: System one support | dhiltgen | XL | No | 116.4h | 73.2h | Oct 03, 2026 |
|
LLM analysis pending This PR has not been analyzed yet. |
||||||
| docs: document image input for decision models | ParthSareen | S | No | 0.6h | 0.6h | Oct 02, 2026 |
SummaryHealth AssessmentAI DetailsTech Stack |
||||||
| docs: link to available decision models | ParthSareen | XS | No | 0.1h | 0.0h | Oct 02, 2026 |
SummaryHealth AssessmentAI DetailsTech Stack |
||||||
| ci: fix missing build context | dhiltgen | XS | No | 0.3h | 0.0h | Oct 02, 2026 |
SummaryHealth AssessmentAI DetailsTech Stack |
||||||
| models: add clef support via llama-server | jmorganca | XL | No | 1.1h | 0.1h | Oct 01, 2026 |
SummaryHealth AssessmentAI DetailsTech Stack |
||||||
| server: report only decision capability for decision models | dhiltgen | S | No | 2.1h | 1.8h | Oct 01, 2026 |
SummaryFeatureReport decision-only capability for models Clients will no longer see decision models in general chat, tools, or thinking, improving user experience by filtering out non-interactive models. Health Assessment
Small
Low
Low
AI Details
Confidence:
0.75
Tech Stack
Languages:
Go
|
||||||
| docs: document System One API | ParthSareen | L | No | 24.8h | 11.7h | Sep 30, 2026 |
SummaryDocsDocument System One API Reference Adds a Decision guide and System One API reference with examples for choice, yes/no, and scoring questions. Documents local availability, request and response fields, and limits. Health Assessment
Large
Low
Low
AI Details
Confidence:
0.10
Tech Stack |
||||||
| create: support explicit model capabilities | dhiltgen | L | No | 1.0h | 1.0h | Sep 29, 2026 |
SummaryFeaturecreate: support explicit model capabilities Adds explicit capability declarations to model files, enabling fine‑grained control over model behavior and improving request scheduling accuracy for System One. Health Assessment
Large
Medium
Low
AI Details
Confidence:
0.20
Tech Stack
Languages:
Go
|
||||||
| llama.cpp: version bump b11232 | dhiltgen | XS | No | 78.0h | 76.0h | Sep 29, 2026 |
SummaryChorellama.cpp: version bump b11232 Updates the llama.cpp dependency to a newer commit, ensuring compatibility and bug fixes. Health Assessment
Small
Low
High
AI DetailsTech Stack |
||||||
| MLX: version bump | dhiltgen | XS | No | 78.5h | 76.5h | Sep 29, 2026 |
SummaryChoreMLX version bump Updates the MLX library version to the latest release. Health Assessment
Small
Low
High
AI Details
Confidence:
0.80
Tech Stack |
||||||
| feat: allow ten web searches per response | ParthSareen | L | No | 144.3h | 0.4h | Sep 28, 2026 |
SummaryFeatureIncrease web search limit to ten per response Allows up to ten web searches per response for Responses and Anthropic, reducing limit errors and improving user experience. Health Assessment
Small
Low
Low
AI Details
Confidence:
0.10
Tech Stack
Languages:
Go
|
||||||
| mlx: bound pull stall retries and let the watchdog interrupt them | dhiltgen | L | No | 97.8h | 97.7h | Sep 28, 2026 |
SummaryBug FixFix MLX download stall handling and retry limits Improves reliability of safetensor model pulls by bounding retries and enabling watchdog to interrupt stalled downloads, preventing infinite loops and ensuring timely failures. Health Assessment
Large
Medium
Medium
AI DetailsTech Stack
Languages:
Go
|
||||||