The clearest signal this week: agent throughput has outgrown development infrastructure designed for humans. Anthropic disclosed that daily merges climbed from roughly 2,500 in May to over 30,000 in September — and promptly built Scion, a version control system to replace Git, plus Antique, a code review system. A seven-person team completed a full monorepo migration in five months. Meta offered a first-hand retrospective on a different path: pushing engineers from human-in-the-loop to human-on-the-loop, letting agents run the full plan→implement→build→test→correct cycle themselves, with verification gates embedded inside the loop. Both cases point to the same conclusion — the bottleneck isn't the model's ability to write code. It's testing, review, and source hosting, all designed around human cadence. The second thread: open-weight models are stratifying by deployment tier. NVIDIA released Nemotron 3.5 in three tiers at once — a 120B multimodal flagship, a 74B-A9B edge version that fits in 24GB of VRAM, and a 4GB embedded version. Reflection AI's Beam uses a 501B-total / 23B-active sparse MoE with 1M context for agentic coding. LiquidAI's LFM2.5-24B-A2B pushes active parameters down to 2B and hits 184 tok/s on an Apple M5 Max. Training-side disclosures landed the same week: NVIDIA published the midtraining recipe that frontier models commonly use but rarely write down. Third: evaluation and reliability are finally being treated as engineering problems rather than leaderboard problems. AWS quantified "silent failures" — monitoring all green while answers quietly go wrong — and showed how a misaligned release window drops pass rates from 78% to 38%. Apple spent $100K in API tokens and 50K GPU hours building RSI-Inference Bench; the best agent scores just 26.7% pass@1 on real vLLM/SGLang tasks. Put those two numbers together and they say more about where agents stand today than any capability claim.
Three papers this week converge on the same underlying question: how should a recommender system allocate compute, supervision signal, and feature assets — by actual contribution rather than nominal scale? TransX decomposes sequence transduction into independently scalable components and decouples training from serving. EWM uses explanations to reconstruct credit assignment for delayed rewards. FORTE turns feature deprecation into an engineering pipeline that can run daily. Different methods, same question: which compute, which signals, which features actually earn their keep. Thread 1: Scaling in generative recommendation shifts from "stack more model" to "allocate compute." LinkedIn's TransX reports a counterintuitive finding — the encoder saturates at moderate depth under fixed memory resolution, and adding more supervision should instead move budget toward candidate-conditioned fusion. Online, it delivers +6.02% CTR and +4.45% conversion while cutting serving cost by nearly 80%. Capacity growth and serving cost can be decoupled. Thread 2: Credit assignment for delayed rewards starts borrowing external help. EWM drops temporal-proximity reward propagation. Instead, a frontier language model generates explanations of "which early events caused the current outcome," and a conservative hierarchical operator recursively redistributes reward back along those attributions. On Crafter and WordCraft, the more delayed the reward, the larger the gain over PPO and LaRe. Thread 3: Feature asset lifecycle becomes its own engineering problem. Pinterest's FORTE removes 43.3% of features from a search relevance model and cuts 29.9% of trainable parameters from a large ranking model, lifting serving throughput 19.8–25.5% with no online A/B regression. For the first time, feature deprecation has an offline pipeline that predicts online outcomes.
Reflection AI open-sourced Beam, a 501B-total/23B-active agentic coding model with 1M effective context under Apache 2.0 — and AWS shipped a full deployment path on day one. On the consumer side, OpenAI pushed ChatGPT Pro to GPT-6 Astra Pro with parallel multi-agent execution, while Perplexity open-
Meta dropped Muse Spark Open, an 816B/42B-active native multimodal MoE that runs locally on two M5 Ultra Mac Studios — the strongest US open-weight release yet. Microsoft unveiled MAI, a fully in-house 3T-param sparse model activating 80B per token. Meanwhile the FTC served civil subpoenas to OpenAI
Anthropic's pre-IPO numbers landed hard: quarterly revenue jumped from $4.8B to $16.8B, with distillation revenue "materially" cut — a rare look at how frontier labs actually make money. On the open-weights side, StepFun dropped Step 5 Preview (600B sparse MoE, 1M context) and a five-person team shi
Google launched Gemini 4 Argon worldwide, a general-purpose physical reasoning model that scores 87% on ERQA embodied reasoning and can teach a humanoid robot to play cornhole from a single video. Databricks hopped on day one, calling it the strongest Gemini it has tested. Meanwhile, David Silver un
Microsoft unveiled its third-gen Maia 300 chip — 2nm chiplets, 288GB HBM4, and a claimed 40% better tokens-per-watt than leading GPU systems, with multi-gigawatt deployment planned for next year. Meta fired back at OpenAI and Anthropic with frontier models Muse Wonder and Muse Wonder Full. Meanwhile
The AI industry's money and compute story sharpened today: Anthropic locked in over $100B/10 years with AWS, OpenAI added $250B in Azure commitments, and CoreWeave sits on ~$104B in backlog — yet none of it is prepaid. Grayscale's Zach Pandl argues cheaper tokens will *raise* compute demand, the cla
One event dominated this week: on October 6, OpenAI published 722 mathematical manuscripts on GitHub. The results came from an undisclosed internal frontier model, spanning algebra, number theory, topology, theoretical computer science, and mathematical logic. Out of 4,000 evaluated problems, it produced results for 90 of the Top 500 open problems across all of mathematics — at an average compute cost of roughly three hours of ChatGPT Pro per result. Princeton's IAS AGMAI issued a formal statement. Anthropic's Levent Alpöge called Result 003, "Quasi-Riemann Hypothesis," "the most important moment in the history of mathematics." At the same time, three retractions, roughly 20% of results being counterexamples, and disputes over Lean verification are all part of the story. If there's one thing worth watching this week, it's not "AI can do math" — it's that "can AI-produced math be verified" became a public argument for the first time. The second thread is how densely the inference stack is adapting to new hardware. NVIDIA Vera Rubin picked up Day 0 support from both vLLM and SGLang within days: vLLM reported MiniMax M3 hitting 7.8x the throughput of GB200 on AgentX, while SGLang reported FP8 MLA speeding up 20% at batch 1 / 128K context and MoE tail fusion cutting 276 kernel launches per step. In the same week, vLLM shipped v0.31.0 (717 commits, 307 contributors) and released the vLLM-Omni technical report for its omni-modal serving runtime. Hardware generations and engine iterations are now meshing on a weekly cadence. The third thread is agent engineering shifting from "can it run" to "can it be governed." GitHub disclosed the impact of agent-scale development on Git infrastructure (monthly commits up 5x year-over-year, pushes up 4.9x). Microsoft turned agent sandboxing into an OS-level Windows primitive (MXC GA). Anthropic launched Claude Managed Agents dynamic workflows in public beta. And several papers directly study agent time budgets, judge reliability, and si
This week's recommender systems research runs along three main threads, with industrial deployment papers and academic work advancing in parallel. Thread 1 — Search pipeline reranking and supervision signal reallocation: Airbnb's SIFT moves filter ranking from hand-crafted ETL features to Transformer sequence modeling, driving +20.0% online filter engagement; Elastic's VLM Teacher supervision study demonstrates that spending teacher signal on negative-sample judgment beats augmenting positives, lifting ViDoRe v2 nDCG@5 from 55.2 to 62.6/63.0. Both point to the same thing — gains in the search pipeline are shifting from "swap the model" to "how to allocate supervision signal." Thread 2 — Operator-level and framework-level improvements for large-scale training and online optimization: Google AI-Hypercomputer's CutBCE pushes multi-label BCE logits out of HBM, speeding up training by 225.9% on 8-chip TPUs; Meta's Population Evolution uses a cross-task shared experiment ledger to lift ranking-task local gains from 7.01% to 8.97%; also from Meta, the billion-scale thumbnail MAB system hands exploration priors to a deep visual quality model at O(B) scale. Thread 3 — Fine-grained preference alignment for generative recommendation: Walmart's QGDPO uses DPO to filter Doc2Query hallucinations and duplicate predictions, eliminating 50% of irrelevant predictions; AIMS converts the deletion intervention effect of a single historical event into ranking margin supervision. Both start from the same angle — training signal is refining from "the overall output is good" to "which part is doing damage."