Tokenstead Weekly - Oct 12, 2026 issue

New guides

Twenty-one ways to run a model locally: the engine catalog

The runtimes catalog now carries 21 verified local inference engines. The pick-by-hardware map: GGUF, Metal, GPU serving, NPU specialists.

/guides/local-inference-engines

China's export draft leaves published weights alone

China's Announcement 48 draft adds 23 AI technologies to export review, but published weights are exempt. What 2T and top-10 thresholds mean for open models.

/guides/china-open-weight-export

Hermes Agent ships on NVIDIA's RTX Spark launch

RTX Spark laptops ship October 16 with Hermes Agent on the launch slide, pickable in Windows' agent setup flow, running Qwen 3.8 27B local on 128GB.

/guides/hermes-rtx-spark

New open-weight models

Muse Wonder

1048k context

/models/muse-wonder

SubQ Snap

384k context

/models/subq-snap

Muse Spark Open 816B-A42B 816.0B

1024k context, 3 quants

/models/muse-spark-open-816b

Cloud prices - current snapshot

Current per-token pricing from the OpenRouter sync. Week-over-week price deltas coming when we add price history.

Kimi K3 - cheapest: Relace $0.25/M input

current per-token pricing snapshot

/models/kimi-k3

DeepSeek V4 Pro - cheapest: Parasail $0.45/M input

current per-token pricing snapshot

/models/deepseek-v4-pro

New rigs this week

4x DGX Spark 15U rack - Niklas 512GB

512GB unified

/rigs/niklas-lenz-spark-rack

Featured this week

8x DGX Spark rack by FPGA des GARÇONS 1024GB

1024GB unified

/rigs/u5mcxttb

Theo's 4x Mac Mini cluster 96GB

96GB unified

/rigs/e6dcoht6