AI Coding Live News

Flower Labs, the company behind the Flower federated learning framework, released Endeavor 1.1 on November 3, its third model this year after the UK sovereign 7B model Lizzy in April and Endeavor 1.0 in September. The update adds image input, an optional EU-only GPU inference tier, and Flower-managed, private-cloud or on-premises deployment, and promises readable thinking traces plus ownership of outputs that customers may use to train their own models. Vendor-reported scores are GPQA 94.4, HumanEval 99.9, AIME 2026 99.9 and MMMU-Pro 81.1, but the comparison uses GPT-5.6 Sol and Claude Fable 5 rather than either lab's newest models, and access is request-only for now.

Zoom

AI data company micro1 released Dual-Arm Sim, which has 10 general-purpose frontier models control a two-arm Franka Panda robot in a simulated operating room built on NVIDIA Isaac Sim, using only camera images and robot sensors, across four tasks: swapping instruments, replacing a faulty drug cartridge, passing a sample between hands, and packing a surgical kit. Only 11 of 800 trials were completed, all by Gemini 4 Argon (9) and GPT-6 Astra (2). Models claimed "task complete" in 213 trials and were wrong every time. All 200 traces of the cartridge-swap task are public on Hugging Face.

Zoom

In an Oct 26 blog post, OpenAI said its internal models produced three new results: a proof of Sogge's local smoothing conjecture in every dimension (which implies the Kakeya and Stein restriction conjectures), a Lean-verified proof that the Exponential Time Hypothesis (ETH) is false, and a rigorous existence proof for four-dimensional QCD and related quantum field theories including pure Yang-Mills. The math and CS papers and Lean proofs are in the openai/math GitHub repo; the physics artifacts are being held back while an external group led by David Gross writes them up. All of this is OpenAI's own claim and still awaits independent verification.

Zoom

Addressing rollout homogenization and zero-variance advantage collapse in group-relative reinforcement learning (GRPO), SERO (arXiv:2610.19397) restores learner-visited states to execute multi-step oracle continuations with policy-reachability weighting—boosting 7B ALFWorld OOD success by +11.20 points to 87.50%—while UNSW's EDEN (arXiv:2610.16105) appends a single anchored Gaussian-perturbed embedding position to a fraction of rollouts, lifting Qwen2.5-Math-7B AIME 2025 Pass@256 by +13.33 points (to 66.67%) without increasing rollout budgets.

Zoom

On 2026-10-15 Meta released weights for Muse Spark Open, a natively multimodal MoE (~816B total / ~42B active per token, 1M context) aimed at long-horizon coding and agents. Weights are on Hugging Face and peers; hosted APIs include Together, Fireworks, Groq and OpenRouter. Reported figures include DeepSWE 60%, MCP-Atlas 89.2% and JobBench 61.8% (see Meta’s evaluation methodology).

Zoom

On 2026-10-15 Meta released Muse Spark Open open weights: an 816B-total / 42B-active MoE with a Perception Encoder, natively multimodal (text/image/video/audio in → text out), 1M context, aimed at coding and long-horizon agents. Weights are on Hugging Face (BF16/NVFP4), SageMaker, Model Garden and Foundry; APIs include OpenRouter, Together, Fireworks and Groq. Meta calls it the most capable US open-weight model to date and, with Apple, optimized it to run locally across two M5 Ultra Mac Studios via MLX.

Zoom

On 2026-10-15 StepFun released open weights for Step 5 Preview as promised: a 600B-total / 27B-active sparse MoE with up to 1M-token context and vision input, aimed at agentic software engineering and knowledge work. Checkpoints are on Hugging Face (FP8, BF16, NVFP4) and ModelScope. Official scores include StepCodeBench 49.0%, FrontierFinance 66.4%, and Artificial Analysis Intelligence Index 44. September coverage was API/index only; this is the first public weight drop.

Zoom

On 2026-10-12 Z.ai introduced GLM-Legal, built on GLM-5.3-Flash with legal mid-training and RL across tens of thousands of legal environments, co-trained with its agent harness. Vendor-reported: 19.2% Harvey LAB all-pass (base 6.7%); LegalBench and PLawBench claimed at parity without scores. No open weights; API by application; AutoClaw support coming.

Zoom

On Oct 9 Microsoft released Microsoft-Decision-1, a decision-scoring model that returns calibrated probabilities over fixed options instead of generating text, aimed at routing, classification, verification, agent guardrails and AI judging. Microsoft reports top accuracy across 36 blind benchmarks, ~35x lower P50 latency than GPT-6 Sol and a 1.3% decision-flip rate under perturbation. Available on Microsoft Foundry and OpenRouter, 32K context, $0.042/M input, free output.

Zoom

Self-evolving search agents construct synthetic training curricula by jointly co-optimizing a question proposer and an answer solver. However, this closed loop frequently triggers a deceptive failure mode termed 'co-cheating': the proposer and solver increasingly converge on shared hallucinations, artificially driving up internal training rewards while external factual correctness degrades. The acclaimed research 'False Frontiers' (arXiv:2609.39102, 330+ upvotes on Hugging Face) diagnoses this structural blind spot and introduces CrossFit. By partitioning source documents into orthogonal subsets and scoring proposals with solvers trained exclusively on complementary splits, CrossFit severs the circular pseudo-label reinforcement loop. Evaluated across seven downstream search benchmarks, CrossFit elevates average scores by 8.8 and 8.4 points on 4B and 9B models, systematically outperforming Search-R1.

Zoom

On-policy self-distillation (OPSD) trains mathematical reasoning models by deploying a privileged teacher that observes ground-truth solutions to supervise student-sampled prefixes. Standard OPSD fixes the teacher parameters across all states, leaving valuable supervision untapped. Researchers from Tencent WeChat and Tencent AI Lab propose Neighborhood OPSD (N-OPSD). They discover that local parameter perturbations in the privileged teacher reveal complementary, reference-aligned corrections across distinct token positions. Offline greedy pruning compiles a compact pool of frozen neighbor experts, while an online MaxPeak and quantile router dynamically selects the optimal expert distribution. Across AIME 2024, AIME 2025, and HMMT 2025, N-OPSD improves Average@12 by up to 2.75 points on Qwen3-1.7B, 4B, and 8B models without any inference overhead.

Zoom