media AI News (smol.ai) · 1h ago · src: 1d ago · releases

Meta's Muse Spark 1.2 and OpenAI's GPT-5.6 Sol updates

from English

Meta announced that its Muse Spark 1.2 model achieved gold-medal performance in five STEM Olympiads and entered the top 5 on the Vals Index at $0.69/test, reportedly 3x cheaper than Kimi. Meta attributed these gains to multi-agent orchestration with parallel reasoning, noting perfect theory scores at APhO and IPhO without external tools.

OpenAI unified its Instant and deep reasoning capabilities into a single paid-chat model powered by GPT-5.6 Sol for Plus/Pro users, introducing a reasoning-effort slider to balance speed and comprehensiveness. The company also expanded free-tier access to unlimited text chats with GPT-5.6 Luna and launched an open standard for Agent Plugins.

The industry focus is shifting toward agentic orchestration and infrastructure, with Cloudflare highlighting Kitesurf for stateless browser automation and Weaviate adding built-in MCP endpoints. Inference routing is becoming a competitive advantage, as seen in Cursor's Router which directs requests to different models based on task type.

Importance 2/3 New feature vs. leaders AI News (smol.ai) Meta AI OpenAI AI agents Reasoning models

Related articles

arxiv arXiv cs.LG · 49d ago · 1 view

LLM-based Hierarchical Control in Multi-Agent Games

A hierarchical system using a pretrained LLM to select RL skill policies outperforms flat RL in a 2v2 King of the Hill environment. It matches hand-crafted behavior tree performance in win rate and is perceived as more human-like by 60% of users, highlighting effective coordination and adaptability without manual rule design.

arxiv arXiv cs.LG · 50d ago

Reverse-Engineering Transformer Attention with Executable Programs

A new method uses program synthesis to generate Python programs that reproduce attention patterns in transformer models. These programs achieve over 75% average Intersection-over-Union similarity on held-out data and can replace up to 25% of attention heads with minimal impact on model performance, increasing perplexity by only 16% on average.

arxiv arXiv cs.LG · 50d ago · 1 view

EfficientRollout: System-Aware Self-Speculative Decoding for RL Rollouts

EfficientRollout introduces a self-speculative decoding framework that reduces rollout and end-to-end latency by up to 19.6% and 12.7% respectively, without compromising final model quality. It uses a quantized drafter derived from the target model and integrates a system-aware toggle policy to avoid compute-bound regimes, enabling effective speculation during evolving policy generations.

arxiv arXiv cs.CL · 50d ago

GraphPO: Graph-based Policy Optimization for Reasoning Models

GraphPO introduces a directed acyclic graph framework to represent reasoning rollouts, merging semantically equivalent paths to reduce redundant exploration. It assigns efficiency and correctness advantages to edges, improving inference efficiency and process supervision while reducing advantage-estimation variance. Experiments show GraphPO outperforms chain- and tree-based methods on three LLMs across reasoning and agentic search tasks under identical token or response budgets.

arxiv arXiv cs.CL · 50d ago · 2 views

Leadership as Coordination Control in Multi-Agent LLM Teams

Process-level coordination control adds value only when the initial majority consensus is unreliable, the task is recoverable, and unguided interaction fails to repair errors. Across multiple models and tasks, no leadership style outperforms others in accuracy, aligning with contingency theory rather than suggesting a failure of the approach.