https://minerka.com/library/agents/ The model is one part. The harness is the tools, the memory, the context, and the check that decide what that model can finish. Mon, 02 Nov 2026 11:29:44 +0000 https://minerka.com/stories/deepseek-dsh-superagent/ https://minerka.com/stories/deepseek-dsh-superagent/ On October 31, 2026, DeepSeek published dsh-superagent-prototype under the MIT license. The README says it is a research prototype for DeepSeek Harness, not an official product: it watches one GitHub repository, opens work items when allowed people ask, and can answer with a pull request. DeepSeek's accompanying architecture note reports a synthetic Super-agent schedule at 12.4 minutes p50 (15.7 p95), and says removing pass-through raised Router cache-read tokens about 37%. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 31, 2026 https://minerka.com/stories/nvidia-nemoclaw/ https://minerka.com/stories/nvidia-nemoclaw/ On October 30, 2026, NVIDIA published NemoClaw open blueprints that combine a tuned agent harness, Nemotron models, NeMo tools, and the OpenShell runtime. NVIDIA says LangChain tuned Deep Agents for Nemotron 3 Ultra and reached near-frontier performance at approximately 10x lower cost per run versus leading closed models. The blueprints currently support OpenClaw, Hermes Agent, and LangChain Deep Agents. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 30, 2026 https://minerka.com/stories/karpathy-hackertube/ https://minerka.com/stories/karpathy-hackertube/ On October 31, 2026, Andrej Karpathy published HackerTube (hackertube.org), a YouTube-style site built with Claude Opus 5.5 in Claude Code. His post says the loop filtered Hacker News comments down to about 40,000 that recommend a YouTube video, then surfaced 6,454 videos across 150 topics from 2007–2026. He puts the build at about two hours; the manual steps he lists are a YouTube API key, a VPS, and the domain. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 31, 2026 https://minerka.com/stories/openai-misalignment-monitors/ https://minerka.com/stories/openai-misalignment-monitors/ On October 29, 2026, OpenAI described an expanded system for monitoring frontier-model agents. In a replay of the Hugging Face incident, OpenAI says the monitors alerted about 30 minutes after the sandbox escape and more than 30 hours before Hugging Face was breached in the real episode. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 29, 2026 https://minerka.com/stories/ai2-novelty-judges/ https://minerka.com/stories/ai2-novelty-judges/ On October 27, 2026, the Allen Institute for AI released a study of models that score whether a research idea is new. For GPT-5.4, one prompt scored 93.5 percent on pairs of human and generated ideas. A shorter prompt on the same pairs scored 40.9 percent. A simple Claude Opus 4.6 prompt cost $2.59 and beat two purpose-built judges that cost $30.66 and $82.05. Start here (ranks, prices, unknown): https://minerka.com/learn/ October 27, 2026 https://minerka.com/stories/lfm3/ https://minerka.com/stories/lfm3/ On October 27, 2026, Liquid AI released LFM3, six models in the 2 billion parameter range on one backbone. Liquid says LFM3-1.7B beats LFM2.5-2.6B with 35 percent fewer parameters. On WebArena-Verified the score is 38.42, against 3.02 for the predecessor. The weights are on Hugging Face. The post does not name the license. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — How models are ranked, and what they cost: https://minerka.com/guide/models/ October 27, 2026 https://minerka.com/stories/genie-4/ https://minerka.com/stories/genie-4/ On October 27, 2026, Google DeepMind announced Genie 4, a world model that generates a scene and its sound together. It is a research preview for trusted testers. DeepMind says Gemini 3.8 Flash watches the frames and checks that each step happened. The post does not give a frame rate or a resolution, and the model is not in the public API. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 27, 2026 https://minerka.com/stories/nemotron-bolt-e4b/ https://minerka.com/stories/nemotron-bolt-e4b/ On October 27, 2026, NVIDIA said Nemotron 3.5 Bolt E4B, the 4 GB model, is available for on-device game characters. NVIDIA says it is about 40 percent higher quality than its prior model on the benchmarks it chose, and that it writes up to about 80 percent fewer tokens than unnamed peers. The context window is 64,000 tokens. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — How models are ranked, and what they cost: https://minerka.com/guide/models/ October 27, 2026 https://minerka.com/stories/gpt-6-1-terra/ https://minerka.com/stories/gpt-6-1-terra/ OpenAI released GPT-6.1 Terra on October 26. Standard API prices are $1 per million input tokens, $0.05 cached, and $5 per million output tokens, which OpenAI says is half the price of GPT-5.6 Terra and GPT-6.1 Sol. On OpenAI's Terminal-Bench 4.0 row, the max-effort score is 50.3 percent, against 21.5 percent for GPT-5.6 Terra. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — How models are ranked, and what they cost: https://minerka.com/guide/models/ October 26, 2026 https://minerka.com/stories/kimi-k3-1-pro/ https://minerka.com/stories/kimi-k3-1-pro/ Moonshot released Kimi K3.1 Pro on October 24, a post-trained Kimi K3 it calls a 3-trillion-class model, with a 1 million token context. On Moonshot's Terminal-Bench 4.0 row it scores 65.7 percent, ahead of GPT-6 Astra at 58.2 and just behind Claude Opus 5.5 at 66.4. The weights are not on the page. Moonshot says they will be released by November 4. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — How models are ranked, and what they cost: https://minerka.com/guide/models/ October 24, 2026 https://minerka.com/stories/epoch-automation-reports/ https://minerka.com/stories/epoch-automation-reports/ Epoch AI scored models on tasks from its own work. Claude Fable 5.1 and GPT-6 Astra each averaged 65 percent of the rubric. Epoch says 100 percent is the line at which a task can be handed over, and that a 65 percent result is still a failed task. Start here (ranks, prices, unknown): https://minerka.com/learn/ October 8, 2026 https://minerka.com/stories/mirendil-base/ https://minerka.com/stories/mirendil-base/ On October 21, 2026, Mirendil said one researcher, working with agents, pretrained Mirendil-Base in 3.5 months for less than $10 million in compute. The post puts the model in the 100 to 500 billion parameter range. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — How models are ranked, and what they cost: https://minerka.com/guide/models/ October 21, 2026 https://minerka.com/stories/claude-duet/ https://minerka.com/stories/claude-duet/ On October 21, 2026, Anthropic opened a waitlist for Claude Duet, an app that messages you and keeps working after you close it. The preview is for Pro and Max subscribers in the US, and Anthropic says it is built around Claude Opus 5.5. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 21, 2026 https://minerka.com/stories/voxtral-3/ https://minerka.com/stories/voxtral-3/ On October 20, 2026, Mistral released Voxtral 3, speech models plus a voice-agent SDK. Voxtral Mini Transcribe 3 is 4 billion parameters at $0.0035 a minute. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — How models are ranked, and what they cost: https://minerka.com/guide/models/ October 20, 2026 https://minerka.com/stories/minimax-m3-1-flash/ https://minerka.com/stories/minimax-m3-1-flash/ The MiniMax-M3.1-Flash card lists about 428 billion parameters, 23 billion active per token, and a 1 million token context. The license name on the card is MiniMax Community. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 17, 2026 https://minerka.com/stories/deepseek-v4-1-flash-1017/ https://minerka.com/stories/deepseek-v4-1-flash-1017/ On October 17, 2026, DeepSeek shipped V4.1-Flash-1017 on its website, app, and API under the name deepseek-flash, and published MIT weights. The card lists 763 billion parameters and a 1 million token context. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — How models are ranked, and what they cost: https://minerka.com/guide/models/ October 17, 2026 https://minerka.com/stories/reflection-beam/ https://minerka.com/stories/reflection-beam/ On October 16, Reflection published Beam's weights. The model has 501 billion parameters, about 23 billion active, and a 1 million token context. The license is Apache 2.0. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 16, 2026 https://minerka.com/stories/muse-spark-open/ https://minerka.com/stories/muse-spark-open/ Meta published the weights for Muse Spark Open on October 15. It has 816 billion parameters, 42 billion active, and a 1 million token context. Meta says two M5 Ultra Mac Studios can hold it. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 15, 2026 https://minerka.com/stories/microsoft-decision-1/ https://minerka.com/stories/microsoft-decision-1/ On October 9, Microsoft released a model that returns a score for each allowed answer instead of writing text. On Microsoft's tests it was the fastest of the set, 35 times quicker than GPT-6 Sol. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 9, 2026 https://minerka.com/stories/muse-wonder/ https://minerka.com/stories/muse-wonder/ Meta launched Muse Wonder and Muse Wonder Full on October 12. Developers can call Muse Wonder now. The fuller model is for Muse Code subscribers. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 12, 2026 https://minerka.com/stories/nemotron-3-5-super-vl/ https://minerka.com/stories/nemotron-3-5-super-vl/ On October 15, 2026, NVIDIA released Nemotron 3.5 Super VL, an open model with 120 billion parameters, and a 24 GB edge model it calls Bolt 74B-A9B. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — How models are ranked, and what they cost: https://minerka.com/guide/models/ October 15, 2026 https://minerka.com/stories/lfm-2-5-24b/ https://minerka.com/stories/lfm-2-5-24b/ On October 13, 2026, Liquid AI released LFM2.5-24B-A2B, a 24 billion parameter model with 2 billion active and a 131,000 token context. The card's license name is LFM 1.0. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 13, 2026 https://minerka.com/stories/agent-lightning/ https://minerka.com/stories/agent-lightning/ On October 7, 2026, Microsoft Research released Agent Lightning v1.0, a training framework of about 3,500 lines that keeps the agent harness used in deployment. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 7, 2026 https://minerka.com/stories/gemini-4-argon-robotics/ https://minerka.com/stories/gemini-4-argon-robotics/ On October 13, 2026, Google posted a robotics note for Gemini 4 Argon. Google says pure reasoning scores 87 percent on Embodied Reasoning QA. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 13, 2026 https://minerka.com/stories/odyssey-3/ https://minerka.com/stories/odyssey-3/ Odyssey released a world model you can move inside, with a public research preview. Start here (test, price, unknown): https://minerka.com/learn/. Full chapter — Agents, loops, and tools: https://minerka.com/guide/frameworks/ October 8, 2026