Open-source agent projects are producing small subagents with verifiable outputs, concise answers, and autonomous task execution. At the same time, research and real-world tests question whether multi-agent collaboration delivers true benefits or only masked fine-tuning, while agents can still ignore instructions.
A home user with two Spark machines and a MacBook Air asks a local LLM agent to decompile and recompile Need for Speed. The agent first suggests DOSBox, then claims decompilation is done, but later reveals it used an open-source engine mimicking the ...
Autark, created by a 14-year-old Austrian developer, turns AI agents into an autonomous team that plans and executes tasks without continuous prompts. The open-source tool supports local models and cloud service CLIs, but removes sandboxing, approval...
A 1.1 GB LLM derived from LFM3-1.7B performs a single search and returns an answer, citation and URL. In tests the base model fails to respond in 62% of live cases, while Obsessive always responds and reduces fabrication. AI-Radar analyzes the trade-...
A fine-tuned LFM3-1.7B acts as a web-search subagent that returns a value, quote, and URL, or "Not found". Trained with LoRA on an RTX 4060 laptop, it cuts invented answers in fixed benchmarks but trades raw live accuracy for reliable replies. Its lo...
A lightweight project shows how to cut verbosity in a local LLM. Oc-e-Non-v1, derived from Qwen3-4B-Instruct-2507, reduces average response length by 50% and bakes concise behavior into the model rather than relying only on prompting. That matters fo...
PLatour shows that part of the gains in PACT systems appears even with non-informative senders or when only the receiver is trained. PACT-Check measures the real contribution of inter-model communication. For on-premise deployments, the difference be...
An open-source framework challenges the PACT premise: passing latent representations between models can improve benchmarks without real information transfer. PLatour adds trainable zero-initialized bridges and PACT-Check, a method to separate genuine...
A single-model multi-agent framework shares one backbone and coordinates GPU-resident KV states through a directed acyclic graph. Zero-copy radix forking, attention-only KV mounting, and packed DAG-GRPO training cut peak KV memory by 2.8x and raise t...
A knowledge-data dual-driven approach updates LLM skill libraries for liability adjudication tasks. KDASO addresses stale skills, semantic execution gaps and context overflow; tests show gains and transfer to similar scenarios.
A Reddit account shows Qwen3.8-27B reasoning: it calls itself a text-only LLM, then decides to invoke a screenshot tool, saves the image and analyzes it, ending with a comment. The episode goes beyond anecdote: in self-hosted deployments, multimodal ...
ByteDance releases Axion 1.0, a 35B agentic LLM with a complete training recipe: two supervised fine-tuning stages followed by reinforcement learning, without mid-training, DPO, or OPD. Beyond weights, it publishes data, pipelines, prompts, and code....
