AI Coding Live News

On 2026-11-05 Hugging Face released the free Ultimate MoE Training Handbook. Co-founder Thomas Wolf cited ~800k GPU-hours, ~70k words and 55 interactive figures; author Nouamane Tazi separately cited ~500k GPU-hours. The Megatron-LM guide covers routing, expert sizing, communication and debugging via the official Space.

Zoom

At its first Forge conference on November 3, open-model inference platform Fireworks launched a batch of products. Ultra-fast, powered by NVIDIA Blackwell, is in public preview at up to 1,500 tokens per second on DeepSeek V4.1 Flash, cutting a 30-turn agent task from 1 minute to 8 seconds. Background Mode runs deferrable work at half price, alongside multi-day prompt caching and higher rate limits. Nexus Desktop brings open-model routing into the Codex app, Claude Desktop and OpenCode, with Pro plans from $8 per month. Fireworks also announced No Eval, serverless LoRA inference, the Fireworks Sovereign private preview and a 60+ partner Frontier Ecosystem.

Zoom

Anthropic's Frontier Red Team published a research post on November 2 and open-sourced FlashProve under Apache 2.0. A model provider commits to hidden weights once, then attaches a zero-knowledge proof to each reply, so a user without the weights can check on an ordinary CPU that the reply really came from that model. That targets routing services that quietly swap in cheaper models and tampered token streams. Anthropic reports about 110 tokens/s on one H100 for Llama 4 Maverick (400B MoE, 4-bit), proofs of about 19 MB per 1,024-token request, and roughly 70-95x the GPU time of vLLM inference. It is an unmaintained research prototype limited to 1,024-token contexts and greedy decoding of an integer version of the model. Anthropic engineer Steve Weis shared the release on X.

Zoom

At Universe 2026 on Oct 28, GitHub made GitHub Agentic Workflows, an open source GitHub Next project, generally available on GitHub Free, Team and Enterprise. Developers describe tasks such as issue triage, CI failure analysis, documentation updates and test writing in natural-language Markdown, and gh aw compiles them into standard Actions YAML run by Copilot or a coding agent of their choice. GA adds self-hosted ARC runners with Docker-in-Docker, a non-experimental gh aw forecast cost estimate, per-workflow model providers and model routing, and Cloud Hypervisor or NVX micro-VM isolation.

Zoom

At GitHub Universe 2026 on Oct 28, GitHub made Copilot Fusion generally available. It merges the multi-model orchestration of the Project HydraFusion research preview with Copilot auto model selection: users pick Copilot Fusion and an effort level (Standard, High or Max) in the model picker, and the system decides which models to use and whether to run a Single, Cascade or Critique workflow. GitHub says Fusion Max scored 60 on the Artificial Analysis Coding Agent Index, two points below Claude Code with Opus 5.5 Max, at about half the API cost per task, about half the tokens and about 40% less time. Each underlying model is billed at its standard rate, and paid plans get 30% off through Nov 30.

Zoom

On Oct 21, Baseten's research team Base Labs released open decision models mammoth-1 (27B) and elephant-1 (9B), full fine-tunes of Qwen3.5 that skip text generation and return per-option probabilities in one forward pass for routing, classification and scoring. Per Baseten, mammoth-1 tops the Jev Decision Index 0.4 at 71.24 and elephant-1 ties for #1 on the Artificial Analysis Decision Index. Weights are on Hugging Face and the Baseten API is free through Oct 27.

Zoom

At PyTorch Conference North America on Oct 20, the CNCF Sandbox project llm-d released version 1.0. It turns inference engines such as vLLM and SGLang into one coordinated serving pool, and 1.0 adds multi-tenant priority and fairness scheduling (flow control), router high availability with zero-downtime rolling upgrades (DisaggregatedSet), a v1 configuration API, and agent-aware session routing with Responses API state. Founded in 2025 by Red Hat, IBM, Google Cloud, CoreWeave and NVIDIA, the project has 500+ contributors, and JetBrains, DigitalOcean and Mistral already run it in production.

Zoom

Modern Mixture-of-Experts architectures increasingly combine fine-grained expert segmentation with shared experts. However, ETH Zurich and HKU researchers identified an 'Exploration Dilemma' (arXiv:2610.22353, GitHub: BingcongLi/MirrorMoE): shared experts homogenize token representations, suppressing routing exploration just as fine-grained segmentation combinatorially enlarges the routing space. MirrorMoE resolves this via token-adaptive suppression of unnecessary experts, cutting pretraining perplexity across all model scales.

Zoom

Three new studies tackle verification, evaluation, and RL alignment for rule-governed generative world models: OmniVerifier-C1 (arXiv:2610.22355, 32 upvotes) uses Decoupled RL to train a decomposed graded visual verifier that reaches 0.684 on ViVerBench and boosts FLUX.1-dev GenEval from 0.68 to 0.79 while enabling Maze and Sudoku world modeling; Meituan's WorldCast benchmark (arXiv:2610.20731, 30 upvotes) evaluates 12 models across 320 multi-step world state prediction cases, finding GPT-Image-2.5 System at 72.8 vs. 42.8 for Nano Banana Pro and 20.1 for the top open model; and NUS's Routing Policy Optimization (RPO, arXiv:2610.22282, 23 upvotes) turns attention routing into an internal exploration action to train video/image diffusion RL 3.1x–4.2x faster than Flow-GRPO and DanceGRPO.

Zoom

Following the release of DeepSeek-V4.1-Flash-1017, vLLM announced Day-0 serving optimization delivering 1.3B tokens per dollar TCO and 830K tok/s per GPU at 100 tok/s/user on SemiAnalysis AgentX (5x agentic throughput gain since Day 0), while OpenRouter launched live multi-provider routing starting at $0.10/M input and $0.50/M output tokens.

Zoom

On 2026-10-14 Inworld AI opened Inworld Inference to every developer, positioning it as an inference service for consumer apps (companions, tutors, voice, interactive fiction). Instead of listing thousands of models it builds a per-model serving stack for open models such as Gemma 4, GLM 5.3 Flash, DeepSeek and MiMo: draft-and-verify (speculative) decoding, a rewritten split-KV verify attention kernel, KV-aware routing and eval-gated low precision. Vendor-reported: Gemma 4 26B p95 first token of 0.31 s, 7.8x faster than the median OpenRouter provider; up to 90% of prompts from cache; 99.999% success over 20M+ GLM 5.3 Flash requests in 30 days. The API is OpenAI-compatible with Flex/Default/Priority/Fast tiers per request.

Zoom

On Oct 12 Subquadratic released SubQ Snap (public preview, subq-snap-preview), a decision model that scores a fixed answer set instead of generating text. Built on Subquadratic Sparse Attention, it reads up to 384K tokens of text and images, returns a decision on 100K tokens in about a second, and—per vendor numbers—scores 61.7 on Decision Index 0.2.1 vs Jev's 57.9. Input costs $0.05/M tokens; output is free.

Zoom

Token-level LLM routing—routing easy tokens to small models and switching to frontier models for critical reasoning tokens—presents an optimal trade-off between output quality and inference cost. However, current LLM serving engines (such as vLLM and TGI) are architected for single-model execution, suffering severe step desynchronization, batch admission latency, and extreme orchestration complexity when handling high-frequency token transfers across heterogeneous models. Researchers from Tsinghua University's NICS Lab present TokenRouter (arXiv:2610.12242, accepted at NeurIPS 2026, code: github.com/thu-nics/TokenRouter), an efficient and developer-friendly serving system engineered for token-level routed LLM inference. TokenRouter introduces a 'request-centric programming and model-centric execution' paradigm, enabling developers to write routing logic intuitively per request while the runtime handles asynchronous execution across subservers. Incorporating a Delayed-Batching Scheduler parameterized via discrete-time Markov chain modeling, a decoupled tri-loop architecture, and an instant handoff-resume mechanism, TokenRouter achieves 2.01x to 64.15x higher decoding throughput compared to existing serving platforms across diverse routing workloads, with complete code open-sourced.

Zoom