AIME

Timeline

1. Oct 18, 2026 · Opinion / analysis · 1 source
   Agens Volundr 32B v1: we fixed what you flagged in the Preview. 0/50 loops, session continuation, and an MIT engine that runs it with 262K context (measured to 239K) within a 16 GB budget
   WHAT v1 FIXES FROM THE PREVIEW
2. Oct 16, 2026 · Open-source release · 5 sources
   Welcome Beam: a 501B open model you can own, customize, and scale
   Beam is focused on coding and agentic use cases.
3. Oct 15, 2026 · Opinion / analysis · 1 source
   QUASAR 4-bit QAT checkpoints for Qwen3.5, Gemma 4 and Muse Glimmer: closer to BF16 than Google’s and NVIDIA’s 4-bit releases
   Two months ago we released a fully quantized NVFP4 Qwen3.8-27B trained with QUASAR.
4. Oct 15, 2026 · Research paper · 1 source
   Router Drift Control: Mitigating Off-Policy Degradation in Mixture-of-Experts Reinforcement Learning
   Reinforcement learning (RL) can become off-policy due to training-inference mismatch, mini-batch updates, and asynchronous training.The resulting mismatch is especially harmful in Mixture-of-Experts (MoE) models, where rollout--training discrepancies can change routing decisions and lead to router drift.
5. Oct 14, 2026 · Opinion / analysis · 1 source
   qwen3.8-flash-next-iq3_s vs -27B-GGUF:UD-Q4_K_XL
   Run with evalscope over the OpenAI-compatible API: Qwen3.8-Next on Strata (MI50 32GB), Qwen3.8-27B on llama.cpp (RX 7900 XTX 24GB).
6. Oct 14, 2026 · Research paper · 2 sources
   PN-Mix: Pair-Norm Guided Mixed-Precision Key Cache Quantization for Efficient Long-Output LLM Inference
   Based on this insight, we propose PN-Mix, a mixed-precision key-cache quantization method for RoPE-based decoder-only LLMs, which allocates higher precision to salient channel pairs identified once during the prefill stage and reused throughout decoding.
7. Oct 13, 2026 · Research paper · 2 sources
   Unlocking Structured Pruning for Reasoning Language Models via Attention-Preserving MLP Trimming
   Structured pruning reduces LLM inference costs by removing model components to produce smaller dense models.
8. Oct 13, 2026 · Research paper · 1 source
   Beyond Answer Sharing: TrueMAD for Targeted Peer Engagement in Multi-Agent Debate
   Multi-agent debate (MAD) aims to improve large language model reasoning by exposing agents to alternative solutions and enabling iterative revision.
9. Oct 12, 2026 · Research paper · 1 source
   Double-Rollout Reinforcement Learning with Sampling-based Reward Estimation
   Reinforcement learning with verifiable rewards (RLVR) implemented with Group Relative Policy Optimization (GRPO) has driven substantial progress in training reasoning language models.
10. Oct 11, 2026 · Research paper · 1 source
    Truncation Samplers Are Nested: Comparing Decoding Rules Is a Measurement Problem
    Settling one comparison between two truncation samplers has cost six thousand A100-hours, and nothing says in advance how far apart two samplers can be.
11. Oct 11, 2026 · Research paper · 1 source
    RECAP: Calibrating Low-Bit Reasoning Models on the History They Read
    We present RECAP, a post-training quantization method that calibrates reasoning models on the history they read.
12. Oct 9, 2026 · Model release · 2 sources
    ConwayResearch/Underdog-Saluki-27B-1.0
    ConwayResearch published the model Underdog-Saluki-27B-1.0 on Hugging Face.
13. Oct 8, 2026 · Research paper · 1 source
    ReCal: Calibrating Structured Pruning for On-Policy Distillation Recovery
    Structured pruning reduces the deployment cost of reasoning language models, but the resulting capability degradation can hinder subsequent on-policy distillation (OPD) recovery.
14. Oct 8, 2026 · Research paper · 1 source
    SFT-as-Context Mitigates Forgetting in Supervised Fine-Tuning
    We introduce SFT-as-context, a training-free method in which the parent model uses the SFT model's response as context to answer the query.
15. Oct 7, 2026 · Research paper · 1 source
    RSIGym: A Flexible Environment for Recursive Self-Improvement
    We introduce RSIGym, an agent-native research environment based on Everything as a Service (EaaS).
16. Oct 7, 2026 · Research paper · 1 source
    YANchor-4B: Effective Long-Horizon Reasoning in O(N) Time with O(1) Memory
    Therefore, we present YANchor-4B, a general-purpose recurrent model that preserves crucial memory as ANchors for retrieval during subsequent reasoning.
17. Oct 7, 2026 · Model release · 2 sources
    [AINews] Reflection Beam - 501B-A23B American Open Model
    It’s been over a year since Reflection launched with us with big goals on coding (and hinted about their RL approach):
18. Oct 7, 2026 · Research paper · 1 source
    BoT-GRPO: Efficient Process-Reward RL for Reasoning via Bag-of-Token Aggregation
    We propose Bag-of-Tokens Group Relative Policy Optimization (BoT-GRPO), which extends GRPO to token-level reward models through a length-invariant "bag of tokens" aggregation: it collects all token-level rewards across rollouts, weights each by the inverse of its source sequence length, and computes per-token advantages relative to weighted group statistics.
19. Oct 7, 2026 · Research paper · 1 source
    From Pareto to Preference: Personalized Test-Time Scaling via Amortized Agentic Policy Discovery
    To reduce the overhead of repeated policy discovery for new user profiles, we propose PersonTTS, an amortized agentic policy-discovery framework that reuses prior search experience through requirement-matched controller initialization and source-distilled procedural guidance, while retaining target-profile evaluation for every candidate.
20. Oct 7, 2026 · Research paper · 1 source
    Collaborative Reasoning Distillation via Cross-Feedback and Coherent Curation
    We propose Collaborative Reasoning Distillation (CRD), a framework that enhances reasoning in compact models through three innovations: (1) interactive cross-feedback where teachers iteratively critique each other's reasoning, (2) fine-grained step-wise quality assessment capturing logical validity independent of final answers, and (3) coherence-aware step stitching that synthesizes complementary strengths.
21. Oct 6, 2026 · Opinion / analysis · 1 source
    Childhood and Education #21: Grades and Standards
    First you get to college via the wrong metrics, as previously discussed in #12.
22. Oct 6, 2026 · Benchmark result · 1 source
    Qwen3.8 Flash Next Reasoning Modes: Off vs Low vs Medium vs Xhigh
    Like Qwen3.8 27B, Qwen3.8 Flash Next has three reasoning efforts: low, medium, and xhigh.

Often appears with