Unverified50% confidenceFactExact time
The core change in Ollama v0.40.5 is porting MLX's Metal custom kernels to the CUDA platform (PR #18657, author @dhiltgen).
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/15/2026
First Seen
Valid until: 1/13/2027
Sources
Related Entities
Related Claims
UnverifiedSGLang v0.5.18 unifies the compiled kernel caches of Triton, FlashInfer, Inductor, DeepGEMM, and the CUDA driver under SGLANG_CACHE_DIR; this is a breaking change.60% similarUnverifiedThe Neo X1 Pro is powered by the AMD Ryzen 9 9955HX3D mobile processor, featuring a 16-core Zen 5 architecture.58% similarUnverifiedOllama v0.40.4 contains three changes directly related to MLX: state compaction after speculative rollback and an MLX / MLX-C version upgrade, both submitted by dhiltgen.57% similarUnverifiedOllama implements cross-platform inference at its core through llama.cpp, supporting three compute backends: NVIDIA CUDA, Apple Metal (M-series chips), and CPU-only.57% similarVerifiedllama.cpp supports GPU acceleration backends including Metal, CUDA, Vulkan, and SYCL.57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1036832API
curl https://kongchang.com/api/v1/knowledge/claims/1036832MCP
get_claim(id=1036832)