All Articles · SURL AI News

Graphify 0.9.66 pins hash seeds to reduce drift in rebuilt code graphs

The new release pins the hash seed used in graph processing and prioritizes directly extracted relationships over inferred results. The patch helps teams compare rebuilds, but model-generated content and consistency across versions still require separate verification.

2026-09-23

GitHub Repo

vLLM 0.30.0 adds resident GPU weight caching for reuse across engine restarts

Fast Start reuses processed weights through CUDA IPC, reducing disk loading during engine restarts. Zero-copy mode is incompatible with sleep mode, and deployers must also validate parallelism configurations and cache fallback behavior.

2026-09-22

GitHub Repo

vLLM Reportedly Terminates Engine After External KV Cache Read Failure; Scope Remains Unclear

A report involving a development build and a hybrid-attention model shows that a single request failure was followed by a fatal scheduler assertion. The documented failure policy should contain failures at the request level, but no public fix or cross-environment validation is available yet.

2026-09-22

Claude Opus 5.5 lowers token prices and introduces multiple breaking changes to the agent API

The new version reduces input, output, and cache read prices and makes adaptive thinking always active. Forced tool calls, edits to conversation history, and streamed progress displays all require renewed validation.

2026-09-22

GitHub Repo

vLLM publishes Qwen3.8-2.4T disaggregated inference benchmarks, reaching about 5,000 total tokens per second per GPU

The team benchmarked separate prefill and decode configurations on GB300 NVL72, providing version details and a configuration repository. Peak throughput and roughly 180 generated tokens per second per user come from different configurations and still require validation under real workloads.

2026-09-22

GitHub Repo

Transformers integrates ggml quantization kernels for direct inference with compressed GGUF weights on Apple Silicon

Hugging Face has announced GGUF inference integration on the main branch, bringing compressed weights into existing Python and PyTorch workflows. Initial support focuses on Apple Silicon and Qwen3.5 architectures, though the benchmarks use different measurement methods.

2026-09-22

GitHub Repo

claude-mem 13.25.3 blocks changes to the Claude CLI executable path over HTTP

The release narrows which settings can be changed through the local settings interface and adjusts the order of telemetry error redaction. Local requests still use the existing trust model, and concerns about redacting quoted credentials remain unverified.

2026-09-22

AI 研究

Shared Inference Cost Study Publishes H100 Results, Showing How Metering Changes Tenants’ Cost Shares

unalloc consolidates cluster and model bills. The study found that different metering methods shifted the same tenant’s share of costs by roughly 12–14 percentage points. Hardware validation consisted of one short run per workload, so the results do not establish a definitive allocation.

2026-09-22

GitHub Repo

Hermes patches Claude CLI plugin so tools with no arguments no longer trigger fallback provider switches

Nous Research updates the DirectSDK plugin’s pinned commit to fix issues with tools that take no arguments, thinking settings, and interface compatibility. The version remains 0.3.0, so deployers must check the commit hash and actual tool execution flow.

2026-09-22

推論系統

NVIDIA publishes multi-GPU video inference benchmarks, cutting generation latency to about 34 seconds on eight GPUs

NVIDIA ran Cosmos 3 Nano using Dynamo-Triton and TensorRT, reducing generation latency from 156.6 seconds on one GPU to 34.2 seconds on eight. The tests did not cover concurrent throughput or per-video costs, and quality validation used only sampled frames.

2026-09-22

GitHub Repo

vLLM Community Proposes RDNA Tiling Optimization, Measuring About 17%–19% Higher Embedding Throughput on W7900

Community tests improved embedding-serving performance on AMD GPUs by reducing the key-value block size in an attention kernel. The change remains a proposal, and kernel speedups, overall throughput, and retrieval quality must be evaluated separately.

2026-09-22

GitHub Repo

Ultralytics 8.4.157 adds CUDA Graph replay, cutting YOLO26n FP16 test latency by about 20%

The release uses real-image calibration and CUDA Graph replay to reduce inference overhead. Gains vary with precision and batch size, while the SiLU rewrite mentioned in the release notes was withdrawn during final review.

2026-09-22