AI Briefing#llm
The latest AI and developer news about #llm, with the original source and a short summary.
The latest AI and developer news about #llm, with the original source and a short summary.
The model is free to use through October 13, 2026.
With improvements in coding, agents, and other areas, it is offered at approximately 1/5 the cost of GPT-6 Astra.
The test generated 1,107 mutation attempts, resulting in 49 validated findings that led to new SSRF detection rules in the Managed Rules update.
The model features 320B total parameters with only 15B activated per token, optimized for reasoning and multi-step tool use.
The release includes DFlash draft models for the 9B, 35B-A3B, and 397B variants of Ornith-1.5.
The 550B-parameter model scored 535.4 out of 600 on the IOI 2026 problem set, surpassing the top human contestant's score of 498.27.
The open-source GitHub Security Lab Taskflow Agent, which requires a GitHub Copilot license, uses custom taskflows to guide LLMs through incremental security auditing steps.
CLM-v0.1-8B, based on Qwen3-8B, released with candidate action selection speed up to 13x faster than Jev
The new Holo4 models, available in 27B dense and 35B-A3B MoE sizes, interact with software across GUI, code, MCP, and API interfaces.
The MIT-licensed curriculum now includes six downloadable volumes and supports eight languages.
Claude Sonnet 5.5 achieves 70.6% on Terminal-Bench 4.0, a massive jump from Sonnet 5's 10.3%, while maintaining the same pricing as its predecessor.
The model supports Zero Data Retention and integrates with the AI SDK, OpenAI-compatible APIs, and Anthropic Messages API.
Self-organizing agent teams achieved 66.7% average accuracy across five benchmarks, surpassing the 59.0% achieved by a perfect router selecting the best individual member.
Benchmarking across 29 datasets shows GLM-5.3-Flash achieves statistical parity with the specialized Jev model for intent classification without fine-tuning, though at higher latency and cost.
The new models accept shared state, named questions, and optional images to return answer distributions in a single forward pass.
Mica v0.1 4B is an Apache-2.0 licensed model that outputs calibrated probabilities for agent decisions, achieving 67.0 accuracy on held-out English sets while maintaining a small local footprint.
Released six open-weight models and a Studio for testing that return the probability of the optimal choice among candidates instead of generating text.
The pipeline generated over 3,000 machine-checked safety and liveness proofs from real-world TLA+ specifications using an agentic prover-reviewer loop.
Canvas allows users to build full-stack applications that run inside the Copilot app and communicate bi-directionally with the AI agent.
The new Swift 1.5 27B model reduces thinking tokens by 58.5% while improving accuracy, with Flash Next offering a 1.8x speed increase.
The new 280M-parameter draft model accelerates LFM2.5-VL-3B inference by up to 3.13x on device and 2.66x on H100 GPUs.
Reflects API structural changes including a 30% output speed improvement over Opus 5, forced thinking activation, and default effort level changes.
The Qwen3-based CLM supports all of Jev's features and achieved 87.6% performance on Terminal-Bench.
CLM-8B achieves SOTA on agent coding benchmarks by separately encoding and caching states and actions.