What changed

---
title: What changed
generated: 2026-10-16
json: https://insidethemachine.ai/changes.json
rss: https://insidethemachine.ai/changes.xml
---

# What changed

What changed in the Inside The Machine data, newest first. Built by scripts/changes.mjs from git diffs of data/*.json and from the news-intake ledger. Every item names its source; nothing is typed in by hand.

Fit rule: fit_tier_gb = smallest standard memory size whose usable share (models.json ram_tiers.usable_gb_assumed) is at least the model's min_memory_gb. Filter new models by fit_tier_gb <= your memory size.

## 2026-10-16: News intake, week of 2026-10-12

Source: data/intake-ledger.json (news packs, merged into the data on 2026-10-16)

- machine-new: Sipeed MaixStation · 2026-10-16 · https://insidethemachine.ai/machines
- model-updated: Nemotron 3.5 Super VL 120B-A12B (needs 105.5 GB, fits 192 GB+ machines) [changed: categories, description, notes, popularity] · 2026-10-16 · https://insidethemachine.ai/models?tier=192#all
- tool-updated: Ollama [changed: seen_in_videos, sources] · 2026-10-16 · https://insidethemachine.ai/setups/ollama
- model-updated: Muse Spark Open 816B-A42B (needs 533 GB, fits ? GB+ machines) [changed: capabilities, categories, context_tokens, description, inputs, links, notes, popularity, release_date, release_date_basis, run] · 2026-10-16 · https://insidethemachine.ai/models#all
- media-model-updated: FLUX.3 Image [klein] [changed: best_for, notes, runs_with, sources, speed_note] · 2026-10-16 · https://insidethemachine.ai/models#images
- model-new: Apertus 1.5 70B (needs 52.3 GB, fits 96 GB+ machines) · 2026-10-15 · https://insidethemachine.ai/models?tier=96#all
- model-new: Apertus 1.5 8B (needs 7.8 GB, fits 16 GB+ machines) · 2026-10-15 · https://insidethemachine.ai/models?tier=16#all
- model-new: UserRepGen 4B (needs 5.1 GB, fits 8 GB+ machines) · 2026-10-15 · https://insidethemachine.ai/models?tier=8#all
- tool-updated: Hermes Agent [changed: sources, stars, strengths] · 2026-10-15 · https://insidethemachine.ai/harnesses
- tool-updated: Hermes Agent [changed: caveats, configure, repo_meta, seen_in_videos, sources, verified_on] ⚠ caveats changed · 2026-10-15 · https://insidethemachine.ai/setups/hermes-agent
- tool-updated: Codex CLI [changed: last_release, sources, stars, strengths] · 2026-10-14 · https://insidethemachine.ai/harnesses
- tool-updated: Ollama [changed: seen_in_videos, sources] · 2026-10-14 · https://insidethemachine.ai/setups/ollama
- model-updated: LFM2.5 24B-A2B (needs 19.2 GB, fits 32 GB+ machines) [changed: benchmarks, categories, description, files, inputs, links, notes, params, popularity, run] · 2026-10-14 · https://insidethemachine.ai/models?tier=32#all
- tool-new: NVIDIA PAIR · 2026-10-13 · https://insidethemachine.ai/setups/nvidia-pair
- model-updated: Nemotron 3.5 Bolt 74B-A9B (needs 30.7 GB, fits 48 GB+ machines) [changed: categories, notes] · 2026-10-13 · https://insidethemachine.ai/models?tier=48#all
- tool-updated: llama.cpp [changed: seen_in_videos, sources] · 2026-10-12 · https://insidethemachine.ai/setups/llama-cpp
- model-new: vLLM Semantic Router d3 27B (needs 24.3 GB, fits 48 GB+ machines) · 2026-10-12 · https://insidethemachine.ai/models?tier=48#all
- tool-updated: Open WebUI [changed: last_release, sources, stars] · 2026-10-12 · https://insidethemachine.ai/harnesses
- tool-updated: Open WebUI [changed: caveats, repo_meta, seen_in_videos, sources, verified_on] ⚠ caveats changed · 2026-10-12 · https://insidethemachine.ai/setups/open-webui
- machine-new: AMD Ryzen AI Halo Developer Platform (Ryzen AI Max+ PRO 495, 192GB) · 2026-10-12 · https://insidethemachine.ai/machines

## 2026-10-16: Site v0.4 → v0.5

Source: git diff f7b15d4..4ad977f -- data/

No data changes. v0.5 changed pages only: agent versions of every page, one memory default, counts read from the data.

## 2026-10-10: News intake, week of 2026-10-05

Source: data/intake-ledger.json (news packs, merged into the data on 2026-10-16)

- model-new: Youtu-Parsing-Omni 5B (needs 5.7 GB, fits 8 GB+ machines) · 2026-10-10 · https://insidethemachine.ai/models?tier=8#all
- machine-new: Microsoft Surface RTX Spark Dev Box (128GB) · 2026-10-10 · https://insidethemachine.ai/machines
- tool-new: GitHub Copilot CLI · 2026-10-10 · https://insidethemachine.ai/harnesses
- tool-new: GitHub Copilot CLI · 2026-10-10 · https://insidethemachine.ai/setups/copilot-cli
- tool-updated: Qwen-Image-2.1-Turbo [changed: hf_checked, license, url_checked, verified_on] · 2026-10-09 · https://insidethemachine.ai/setups/qwen-image-2-1-turbo
- media-model-updated: Qwen-Image 2.1 Turbo [changed: notes, sources] · 2026-10-09 · https://insidethemachine.ai/models#images
- tool-updated: LM Studio Bionic [changed: sources, strengths, weaknesses] · 2026-10-09 · https://insidethemachine.ai/harnesses
- tool-updated: LM Studio Bionic [changed: tags, url_checked, verified_on] · 2026-10-09 · https://insidethemachine.ai/setups/lm-studio-bionic
- machine-new: iBUYPOWER Pro AMD AI Halo Mini PC (Ryzen AI Max+ PRO 495) · 2026-10-09 · https://insidethemachine.ai/machines
- model-new: Mellum2.1 12B-A2.5B Thinking (needs 11.3 GB, fits 16 GB+ machines) · 2026-10-08 · https://insidethemachine.ai/models?tier=16#all
- tool-updated: ML Drift [changed: repo_meta, tags, verified_on] · 2026-10-08 · https://insidethemachine.ai/setups/ml-drift
- machine-new: ASUS ProArt GR1X (NVIDIA RTX Spark) · 2026-10-08 · https://insidethemachine.ai/machines
- tool-updated: Ollama [changed: caveats, repo_meta, verified_on] ⚠ caveats changed · 2026-10-07 · https://insidethemachine.ai/setups/ollama
- model-updated: Mistral Large 4 (needs 726.5 GB, fits ? GB+ machines) [changed: notes] · 2026-10-07 · https://insidethemachine.ai/models#all
- tool-updated: EmbeddingGemma 2 [changed: hf_checked, url_checked, verified_on] · 2026-10-07 · https://insidethemachine.ai/setups/embeddinggemma-2
- model-updated: EmbeddingGemma 2 (270M-740M) (needs 3.5 GB, fits 8 GB+ machines) [changed: benchmarks, notes, popularity] · 2026-10-07 · https://insidethemachine.ai/models?tier=8#all
- machine-new: Gigabyte W775-V10-L01 (DGX Station GB300) · 2026-10-07 · https://insidethemachine.ai/machines
- model-new: Beam 1.0 501B A23B (needs 347.9 GB, fits 512 GB+ machines) · 2026-10-06 · https://insidethemachine.ai/models?tier=512#all
- model-updated: GLM-5.3-Flash (needs 233 GB, fits 512 GB+ machines) [changed: notes] · 2026-10-06 · https://insidethemachine.ai/models?tier=512#all
- tool-updated: llama.cpp [changed: caveats, repo_meta, sources, verified_on] ⚠ caveats changed · 2026-10-06 · https://insidethemachine.ai/setups/llama-cpp
- machine-new: Minisforum MS-S1 MAX-P495 (Ryzen AI Max+ PRO 495 "Gorgon Halo") · 2026-10-06 · https://insidethemachine.ai/machines
- tool-updated: ThinkingBox [changed: repo_meta, verified_on] · 2026-10-05 · https://insidethemachine.ai/setups/thinkingbox
- model-new: Kolibri 1 (needs 56.6 GB, fits 96 GB+ machines) · 2026-10-05 · https://insidethemachine.ai/models?tier=96#all
- model-new: AstaBrief 8B (needs 7.8 GB, fits 16 GB+ machines) · 2026-10-05 · https://insidethemachine.ai/models?tier=16#all

## 2026-10-02: News intake, week of 2026-09-28

Source: data/intake-ledger.json (news packs, merged into the data on 2026-10-16)

- machine-updated: DGX Spark 64GB (partner GB10 systems) [changed: notes] · 2026-10-02 · https://insidethemachine.ai/machines
- tool-updated: Pi (pi-mono) [changed: sources, stars] · 2026-10-02 · https://insidethemachine.ai/harnesses
- tool-updated: Pi (pi-mono) [changed: homepage, repo_meta, verified_on] · 2026-10-02 · https://insidethemachine.ai/setups/pi-pi-mono
- tool-updated: Clef [changed: hf_checked, url_checked, verified_on, what] · 2026-10-01 · https://insidethemachine.ai/setups/cloudflare-clef
- tool-updated: Strands Decider [changed: repo_meta, verified_on] · 2026-10-01 · https://insidethemachine.ai/setups/strands-decider

## Dropped from the directory (date not recorded)

- huggingface/text-generation-inference: archived
- nomic-ai/gpt4all: stale: last push 2025-05-27
- huggingface/huggingface_hub: off-topic / not local-AI specific
- modelscope/modelscope: off-topic / not local-AI specific
- RooCodeInc/Roo-Code: archived
- plandex-ai/plandex: stale: last push 2025-10-03
- zed-industries/zed: off-topic / not local-AI specific
- voideditor/void: archived
- Doriandarko/claude-engineer: stale: last push 2024-12-12
- MoonshotAI/kimi-cli: archived
- github/copilot-cli: off-topic / not local-AI specific
- ollama/ollama-python: off-topic / not local-AI specific
- langchain-ai/langchain: off-topic / not local-AI specific
- campfirein/byterover-cli: archived
- stanford-oval/storm: stale: last push 2025-09-30
- AnswerDotAI/RAGatouille: stale: last push 2025-05-17
- getomni-ai/zerox: stale: last push 2025-05-20
- lm-sys/RouteLLM: stale: last push 2024-08-10
- Not-Diamond/RoRF: stale: last push 2024-09-24
- tensorzero/tensorzero: archived
- kevinhermawan/Ollamac: stale: last push 2025-03-12
- ollama/ollama-js: off-topic / not local-AI specific
- nomic-ai/gpt4all-chat: archived
- hexgrad/kokoro: stale: last push 2025-08-06
- rhasspy/piper: archived
- coqui-ai/TTS: stale: last push 2024-08-16
- suno-ai/bark: stale: last push 2024-08-19
- myshell-ai/OpenVoice: stale: last push 2025-04-19
- lllyasviel/stable-diffusion-webui-forge: stale: last push 2025-07-31
- argmaxinc/DiffusionKit: archived
- black-forest-labs/flux: stale: last push 2025-07-31
- axolotl-ai-cloud/axolotl-cookbook: off-topic / not local-AI specific
- casper-hansen/AutoAWQ: archived
- Cornell-RelaxML/quip-sharp: stale: last push 2024-10-29
- mit-han-lab/llm-awq: stale: last push 2025-07-17
- LiveCodeBench/LiveCodeBench: stale: last push 2025-07-16
- lmarena/arena-hard-auto: stale: last push 2025-06-21
- tatsu-lab/alpaca_eval: stale: last push 2025-08-09
- Aider-AI/polyglot-benchmark: stale: last push 2024-12-22
- ray-project/llmperf: archived
- aj-geddes/claude-code-bmad-skills: low adoption (487 stars)
- Aider-AI/conventions: stale: last push 2025-06-25
- tlkh/asitop: stale: last push 2024-04-18
- aristocratos/btop: off-topic / not local-AI specific
- ml-explore/mlx-data: off-topic / not local-AI specific
- huggingface/accelerate: off-topic / not local-AI specific
- intel/ipex-llm: archived
- ivanfioravanti/llm_context_benchmarks: off-topic / not local-AI specific
- abiosoft/colima: off-topic / not local-AI specific
- lima-vm/lima: off-topic / not local-AI specific
- tailscale/tailscale: off-topic / not local-AI specific
- vllm-project/vllm: trimmed for focus: category cap (93863 stars); verified live, can be re-added
- mudler/LocalAI: trimmed for focus: category cap (49555 stars); verified live, can be re-added
- exo-explore/exo: trimmed for focus: category cap (47907 stars); verified live, can be re-added
- microsoft/BitNet: trimmed for focus: category cap (41169 stars); verified live, can be re-added
- huggingface/candle: trimmed for focus: category cap (21172 stars); verified live, can be re-added
- ggml-org/ggml: trimmed for focus: category cap (15484 stars); verified live, can be re-added
- NVIDIA/TensorRT-LLM: trimmed for focus: category cap (14822 stars); verified live, can be re-added
- abetlen/llama-cpp-python: trimmed for focus: category cap (10648 stars); verified live, can be re-added
- InternLM/lmdeploy: trimmed for focus: category cap (8105 stars); verified live, can be re-added
- google-ai-edge/LiteRT-LM: trimmed for focus: category cap (6679 stars); verified live, can be re-added
- pytorch/executorch: trimmed for focus: category cap (5095 stars); verified live, can be re-added
- turboderp-org/exllamav2: trimmed for focus: category cap (4635 stars); verified live, can be re-added
- zml/zml: trimmed for focus: category cap (4180 stars); verified live, can be re-added
- google-ai-edge/LiteRT: trimmed for focus: category cap (3497 stars); verified live, can be re-added
- ikawrakow/ik_llama.cpp: trimmed for focus: category cap (3304 stars); verified live, can be re-added
- b4rtaz/distributed-llama: trimmed for focus: category cap (3109 stars); verified live, can be re-added
- ml-explore/mlx-swift-examples: trimmed for focus: category cap (2682 stars); verified live, can be re-added
- ml-explore/mlx-swift: trimmed for focus: category cap (2059 stars); verified live, can be re-added
- microsoft/onnxruntime-genai: trimmed for focus: category cap (1137 stars); verified live, can be re-added
- madroidmaq/mlx-omni-server: trimmed for focus: category cap (745 stars); verified live, can be re-added
- SWE-agent/SWE-agent: trimmed for focus: category cap (20573 stars); verified live, can be re-added
- twinnydotdev/twinny: trimmed for focus: category cap (3670 stars); verified live, can be re-added
- agentskills/agentskills: trimmed for focus: category cap (26178 stars); verified live, can be re-added
- VoltAgent/awesome-claude-code-subagents: trimmed for focus: category cap (25742 stars); verified live, can be re-added
- SuperClaude-Org/SuperClaude_Framework: trimmed for focus: category cap (23909 stars); verified live, can be re-added
- agent0ai/agent-zero: trimmed for focus: category cap (19481 stars); verified live, can be re-added
- leon-ai/leon: trimmed for focus: category cap (17578 stars); verified live, can be re-added
- nanobrowser/nanobrowser: trimmed for focus: category cap (14097 stars); verified live, can be re-added
- simular-ai/Agent-S: trimmed for focus: category cap (12623 stars); verified live, can be re-added
- e2b-dev/open-computer-use: trimmed for focus: category cap (2363 stars); verified live, can be re-added
- camel-ai/camel: trimmed for focus: category cap (17889 stars); verified live, can be re-added
- microsoft/agent-framework: trimmed for focus: category cap (14145 stars); verified live, can be re-added
- strands-agents/harness-sdk: trimmed for focus: category cap (8824 stars); verified live, can be re-added
- zylon-ai/private-gpt: trimmed for focus: category cap (57582 stars); verified live, can be re-added
- run-llama/llama_index: trimmed for focus: category cap (52512 stars); verified live, can be re-added
- milvus-io/milvus: trimmed for focus: category cap (46389 stars); verified live, can be re-added
- microsoft/graphrag: trimmed for focus: category cap (36348 stars); verified live, can be re-added
- PromtEngineer/localGPT: trimmed for focus: category cap (22193 stars); verified live, can be re-added
- allenai/olmocr: trimmed for focus: category cap (19747 stars); verified live, can be re-added
- huggingface/sentence-transformers: trimmed for focus: category cap (19177 stars); verified live, can be re-added
- weaviate/weaviate: trimmed for focus: category cap (16885 stars); verified live, can be re-added
- Unstructured-IO/unstructured: trimmed for focus: category cap (15569 stars); verified live, can be re-added
- neuml/txtai: trimmed for focus: category cap (13017 stars); verified live, can be re-added
- xberg-io/xberg: trimmed for focus: category cap (9418 stars); verified live, can be re-added
- huggingface/text-embeddings-inference: trimmed for focus: category cap (5079 stars); verified live, can be re-added
- michaelfeil/infinity: trimmed for focus: category cap (2957 stars); verified live, can be re-added
- PrefectHQ/fastmcp: trimmed for focus: category cap (28088 stars); verified live, can be re-added
- modelcontextprotocol/typescript-sdk: trimmed for focus: category cap (13586 stars); verified live, can be re-added
- BeehiveInnovations/pal-mcp-server: trimmed for focus: category cap (11782 stars); verified live, can be re-added
- mcp-use/mcp-use: trimmed for focus: category cap (10770 stars); verified live, can be re-added
- lastmile-ai/mcp-agent: trimmed for focus: category cap (8573 stars); verified live, can be re-added
- modelcontextprotocol/registry: trimmed for focus: category cap (7365 stars); verified live, can be re-added
- executeautomation/mcp-playwright: trimmed for focus: category cap (5666 stars); verified live, can be re-added
- IBM/mcp-context-forge: trimmed for focus: category cap (4640 stars); verified live, can be re-added
- metatool-ai/metamcp: trimmed for focus: category cap (2707 stars); verified live, can be re-added
- docker/mcp-gateway: trimmed for focus: category cap (1601 stars); verified live, can be re-added
- ihor-sokoliuk/mcp-searxng: trimmed for focus: category cap (1307 stars); verified live, can be re-added
- alexrozanski/LlamaChat: trimmed for focus: category cap (1506 stars); verified live, can be re-added
- pipecat-ai/pipecat: trimmed for focus: category cap (16509 stars); verified live, can be re-added
- SWivid/F5-TTS: trimmed for focus: category cap (15409 stars); verified live, can be re-added
- k2-fsa/sherpa-onnx: trimmed for focus: category cap (15302 stars); verified live, can be re-added
- livekit/agents: trimmed for focus: category cap (14757 stars); verified live, can be re-added
- huggingface/speech-to-speech: trimmed for focus: category cap (13471 stars); verified live, can be re-added
- kyutai-labs/moshi: trimmed for focus: category cap (11246 stars); verified live, can be re-added
- KoljaB/RealtimeSTT: trimmed for focus: category cap (10190 stars); verified live, can be re-added
- OHF-Voice/piper1-gpl: trimmed for focus: category cap (5884 stars); verified live, can be re-added
- remsky/Kokoro-FastAPI: trimmed for focus: category cap (5552 stars); verified live, can be re-added
- KoljaB/RealtimeTTS: trimmed for focus: category cap (4042 stars); verified live, can be re-added
- Purfview/whisper-standalone-win: trimmed for focus: category cap (3216 stars); verified live, can be re-added
- idiap/coqui-ai-TTS: trimmed for focus: category cap (2337 stars); verified live, can be re-added
- Tencent-Hunyuan/HunyuanVideo: trimmed for focus: category cap (12623 stars); verified live, can be re-added
- ostris/ai-toolkit: trimmed for focus: category cap (12310 stars); verified live, can be re-added
- Lightricks/LTX-Video: trimmed for focus: category cap (11108 stars); verified live, can be re-added
- Acly/krita-ai-diffusion: trimmed for focus: category cap (10725 stars); verified live, can be re-added
- deepbeepmeep/WanGP: trimmed for focus: category cap (10536 stars); verified live, can be re-added
- QwenLM/Qwen-Image: trimmed for focus: category cap (8407 stars); verified live, can be re-added
- kohya-ss/sd-scripts: trimmed for focus: category cap (7244 stars); verified live, can be re-added
- mcmonkeyprojects/SwarmUI: trimmed for focus: category cap (4669 stars); verified live, can be re-added
- nunchux-ai/nunchaku: trimmed for focus: category cap (3960 stars); verified live, can be re-added
- meta-pytorch/torchtune: trimmed for focus: category cap (5817 stars); verified live, can be re-added
- huggingface/autotrain-advanced: trimmed for focus: category cap (4613 stars); verified live, can be re-added
- sierra-research/tau2-bench: trimmed for focus: category cap (2240 stars); verified live, can be re-added
- sierra-research/tau-bench: trimmed for focus: category cap (1464 stars); verified live, can be re-added
- podman-container-tools/podman: off-topic / not local-AI specific
- ROCm/legacy-rocm-build: repo moved to a legacy build repo; ROCm docs live elsewhere
- ml-explore/mlx-community: not found on GitHub (404/renamed without redirect)
- ggml-org/gguf-my-repo: not found on GitHub (404/renamed without redirect)
- msty-app/msty: not found on GitHub (404/renamed without redirect)
- deep-chat/deep-chat: not found on GitHub (404/renamed without redirect)
- ml-explore/mlx-whisper: not found on GitHub (404/renamed without redirect)
- NousResearch/hermes-agent-skills: not found on GitHub (404/renamed without redirect)
- lunary-ai/lunary: not found on GitHub (404/renamed without redirect)
- nvtop/nvtop: not found on GitHub (404/renamed without redirect)
- ollama/ollama-tool: not found on GitHub (404/renamed without redirect)
- undefined: cloud service, no public local tool
- undefined: cloud-only model
- undefined: weights not released yet (watchlist)
- undefined: weights not released yet (watchlist)
- undefined: unconfirmed report, no source
- undefined: closed model, restricted access
- undefined: cloud service
- undefined: hardware: handed to /machines