Infrastructure & Chips
-
Open Source
SwarmLLM splits a 27B model across browser tabs
SwarmLLM v0.2.0 splits a 27B model across browser tabs with WebGPU and WebRTC, trading single-device memory limits for network and peer-trust constraints.
-
Research
Prefix-cache state can make quantized agent runs diverge at temperature zero
A new reproducibility study finds cache-state differences can alter agent trajectories even at temperature zero, with larger effects under weight quantization.
-
New Products
NVIDIA PAIR scales local agent inference by routing requests, not pooling VRAM
NVIDIA PAIR can widen local agent throughput across PCs and Macs, but it routes independent requests rather than pooling VRAM or sharding models.
-
Open Source
Soup v0.74 fixes an fp32 load path that inflated LoRA GPU memory
Soup v0.74.0 fixes frozen-base fp32 loading, reports a large scoped H100 memory reduction, and tightens serving security controls.
-
Research
Uno turns diffusion into a lossless speed layer for autoregressive LLMs
Uno combines autoregressive LLMs with diffusion adapters and Ψ-Spec verification, releasing code and checkpoints for parallel, distribution-preserving decoding.
-
Agents
Zoho Catalyst 3.0 gives coding agents controlled access to cloud infrastructure
Catalyst 3.0 connects coding agents to cloud infrastructure through MCP, with scoped permissions, audit logs and reversible changes.
-
Open Source
FreeToken roadmap targets AMD, Apple Silicon and multi-GPU expansion
FreeToken's 2026 roadmap targets AMD ROCm, Apple Silicon, DGX Spark, multi-GPU, multimodal models and speculative decoding beyond its current NVIDIA-only support.
-
Infrastructure & Chips
EuroHPC locks in EUR 387.8M for LUMI-AI public compute capacity
LUMI-AI will add AMD-powered public AI compute for European startups, SMEs and researchers, with multi-tenant architecture and API access planned for 2027.
-
New Products
Perplexity moves its Computer agent onto local NVIDIA hardware
Perplexity Portable Computer runs agent models, orchestration, local search and sandboxed tools on NVIDIA DGX Spark with optional cloud escalation.
-
AI data centers are moving toward power, not just users
JLL data suggests new European hyperscale projects are moving farther from major hubs as power, land and grid access reshape AI infrastructure geography.