Search “Reasoning”

Metis Reasoning is founded in London, focusing on a compute-efficient AI approach

On October 20, 2026, the London-based AI company Metis Reasoning was officially established and launched, focusing on a compute-efficient reasoning approach as it seeks a direction distinct from the frontier models of Anthropic and OpenAI. According to Tech.eu, the company was co-founded by Thore Graepel, former head of research at Google DeepMind and co-creator of AlphaGo, and has a team of 12 people spanning researchers, mathematicians, robotics experts and technology entrepreneurs. The company believes that LLMs have fundamental limitations in reasoning, with results that are difficult to verify and lacking an audit trail; it says its approach is more verifiable, more robust and safer, and that it has already achieved some positive results on frontier reasoning problems at a cost far below that of frontier models. The report says its initial funding is believed to be in the tens of millions of dollars, but Metis has not yet announced a release date for its first AI model.

Fireworks launches Free Weights Club and opens free SFT training for Muse Glimmer 30B

Fireworks AI launches the Free Weights Club and opens free SFT training for Muse Glimmer 30B. According to its announcement, the SFT training is free until November 30, with no credit card required, while inference and evaluation calls are billed separately. The same source also says that a tuned Glimmer trained on Kalshi market data achieved 58.3% prediction accuracy on unseen topics, 11.2 percentage points higher than Jev's 47.1%; in a historical backtest, a hypothetical $1,000 grew to $1,071 after simulated fees, while both base Glimmer and Jev posted losses. In another synthetic benchmark, the System One serving stack had a median server-side processing speed 2-3 times faster than standard BF16 chat completions. There are currently no follow-up developments.

doubleAI releases its first code execution model, Maxwell

On October 19, 2026, Amnon Shashua's new company doubleAI released its first product, Maxwell, describing it as belonging to a new category called Code Execution Models (CEM), which optimizes software execution efficiency between code and chips without modifying source code. According to CTech, doubleAI says Maxwell continuously analyzes running programs, hardware, and data and optimizes the underlying machine instructions, while using formal mathematical verification to ensure that each optimization does not change the program's original behavior, with support for CPUs and GPUs. The company says Maxwell speeds up data pipelines by more than 38x and AI inference kernels by more than 4.5x on real workloads; these figures have not been independently verified. Maxwell is now available for early access through the official website, and doubleAI plans to charge customers based on the computing cost saved. There are currently no subsequent development facts.

TinyFish makes Search and Fetch free

TinyFish announced that it will waive retrieval fees for Search and Fetch within the limits. According to a MarkTechPost report, Search's free limit is 30 requests per minute and 500 per hour, while Fetch's is 150 URLs per minute and 1,000 per day. The report also notes that free retrieval does not eliminate the inference costs of downstream models: the same article of about 4,300 characters returns an estimated 1,184 to 42,774 tokens across different providers, and three pages of unfiltered content could fill a 128K context window. In the September 28, 2026 Artificial Analysis snapshot, TinyFish Search scored 71 at $9.30 per 1,000 tasks, the lowest total cost among the displayed configurations but also the lowest score, with each task taking about 59 seconds. There are currently no subsequent developments.

The author announced that the practical experience had been compiled into "Eight Lectures on Agent Engineering in Practice"

On October 18, 2026, the author announced that the practical experience had been compiled into "Eight Lectures on Agent Engineering in Practice". On the same day, the Tencent Cloud Developer Community published an introduction to the series, describing how it focuses on upgrading prototypes that can call models and tools into systems that can complete business tasks, explain failures, control costs, and iterate continuously. The introduction states that the entire series runs through two fictional teaching cases—a case-filing assistant Agent and a postoperative follow-up Agent—covering eight topics: task contracts, reasoning and collaboration, context and memory, reliable tool execution, observability and replay, evaluation and calibration, security boundaries, and production operations; the text emphasizes that a model being able to suggest the next step does not mean the system can deliver business results, and that the example fields, budgets, and figures are all teaching calculations and do not constitute legal advice. Currently, only the introduction to the series has been published, and no factual evidence has yet been seen for the content of the subsequent lectures.

Shao Meng introduces Lecture 10 of Stanford CS336, Inference

On October 18, 2026, Shao Meng published an article introducing Lecture 10, "Inference," of Stanford University's CS336 course. According to the article, this lecture was taught by Percy Liang and uses a single mathematical thread to connect inference optimization at three levels—architecture, algorithms, and systems: at the architecture level, GQA/MLA/CLA reduce dimensionality, local attention truncation, quantization, and pruning and distillation compress the KV cache and parameters; at the algorithm level, speculative sampling achieves lossless acceleration; at the systems level, continuous batching, paging, prefix sharing, and copy-on-write are introduced. The article also notes that the lecture notes are in executable form, with formulas derived step by step by sympy and verified by assertions; Percy Liang believes that the KV cache is the structural root of the Transformer's unfriendliness to inference, and that architectures natively designed for inference (SSMs, linear attention, diffusion-based non-autoregressive generation) may offer the greatest room for improvement. Currently, this content has only one publication record, from Shao Meng.

Microsoft and City University of Hong Kong propose Taste-Bench to evaluate agent decision-making

Researchers from Microsoft and City University of Hong Kong have built Taste-Bench to evaluate the long-term decision-making capabilities of AI agents on long tasks. The benchmark contains 502 binary-choice questions. According to The Decoder, the best-performing model, GPT-5.6 Sol, achieved an average accuracy of only 59.7%, while GPT-5.5 reached 59.5%. When key clues only become apparent after a large amount of subsequent work, the average accuracy of all models drops to 21%, close to random guessing, and increasing the reasoning budget brings almost no improvement. The researchers also distilled the teacher model's judgments using Qwen3.6-27B as the base model; the student's accuracy on unseen tasks rose from 30% to 47.9%, and its success rate on 41 SWE-bench Pro tasks increased from 14.6% to 33.7%.

Perplexity Decider launches on Vercel AI Gateway

On October 17, 2026, Perplexity's Decider V1.2 decision model launched on Vercel AI Gateway. Vercel developers can now use perplexity/pplx-decider-v1.2 for complex agentic workflows and perplexity/pplx-decider-flash-v1.2 for ultra-fast decisions. Aravind Srinivas reposted the news. Earlier the same day, reports said Perplexity had open-sourced two decision models, pplx-decider-flash-v1.2 and pplx-decider-v1.2: the former has 4B parameters and costs 1 cent per million tokens, and is called the strongest 4B model on the HF Decision Index; the latter has 27B parameters and costs 2 cents per million tokens, and ranks 2nd on the overall HF Decision Index leaderboard. Both are now available on the Perplexity Decisions API and offer open weights on Hugging Face. Perplexity Developers said this generation of open-weight multimodal decision models improves both text and vision scores, delivers inference up to 5 times faster, and completes all questions in a single request in one forward pass.

Hao AI Lab evaluates detective agent teams with 200 cases

Hao AI Lab evaluated three detective agent configurations on 200 Victorian-era London cases: the Jev cluster acting alone, GPT6-Astra acting alone, and GPT6-Astra leading the Jev cluster. The institution's published conclusion is that the cluster can cover the clues, but a reasoning leader is needed to spot contradictions and pursue them in depth. In 100 harder cases, with more paid witnesses than honest witnesses, GPT6-Astra plus the Jev cluster solved 93, GPT6-Astra alone solved 59, the Jev cluster alone solved 42, and a rule-based leader plus the Jev cluster solved only 0–1. There are currently no follow-up developments.

Hao AI Lab announced the results of its multi-agent crime-solving benchmark

On October 17, 2026, Hao AI Lab announced the results of its multi-agent crime-solving benchmark, which used 200 Victorian London detective cases. Among 100 harder cases in which paid witnesses outnumbered honest ones, GPT6-Astra leading the Jev cluster solved 93 cases, GPT6-Astra alone solved 59 cases, the Jev cluster alone solved 42 cases, and a rule-based leader plus the Jev cluster solved only 0–1 cases. Hao AI Lab also noted that when working alone, each of the two approaches has its own failure mode: the leaderless Jev swarm is like a rumor mill—one child misreads a clue, its companions copy the vote, and the whole street accepts the wrong answer; while the lone genius is limited by time and cannot deliver the decisive clue to the door. The authors conclude that, in the face of organized lies, a reasoning leader is needed who will ask, "Who is speaking?"

Nadella says new research shows machine learning can improve 2- to 6-week weather forecasts

Satya Nadella introduced new research showing that machine learning can improve weather forecasts two to six weeks ahead, a time scale that has historically been difficult to predict. The research has been published in Nature. Nadella said the work could help respond to extreme weather, manage natural resources, and support agricultural decision-making. He also offered a broader idea: the future lies not only in more capable models, but in building intelligent verification systems around models—learning where models systematically err, reasoning about causal relationships, and continuously calibrating predictions against reality. At present, this news has only this one source, and there are no subsequent developments.

OpenRouter launches RespanAI decision model Span-02

OpenRouter announced the launch of RespanAI's open-weight decision model Span-02. The model accepts text and image inputs and returns options, yes/no/unknown, or scores with probabilities in response to custom questions; pricing is $0.01 per million input tokens with free output, and the context length is 131K. OpenRouter later clarified that Span-02 runs on the OpenRouter Decisions API, uses the same request format as other decision models, and switching requires only changing one line with the model name; images and text are placed in the same state, with a maximum of 4 images per request. Respan previously introduced Span-02 as supporting anytime inference, returning the current best decision once a deadline is given, allowing configurable reasoning duration, and supporting multimodal viewing of images and screens. The model is now live on OpenRouter.

English translation is being prepared.

Utho Cloud Accepts B300 GPU Capacity Reservations in India

Utho Cloud has opened reservations for NVIDIA B300 GPU capacity, with local deployment in India planned for February 2027. According to Analytics India Magazine, the service offers rupee billing and local Indian hosting, and customers can reserve compute for 1, 2, or 3 years, with allocations made in reservation order. The B300 is based on the Blackwell Ultra architecture, features 288GB of HBM3e memory, and targets large language model training, inference, and AI agent workloads. There are currently no public facts about further deployment progress.

English translation is being prepared.

English translation is being prepared.

Google opens Gemini 4 Argon to developers

On October 13, 2026, Google announced that developers can now use Gemini 4 Argon to build applications. This frontier model targets complex workflows such as real software engineering, enterprise knowledge work, and cybersecurity defense, and is designed for long-horizon tasks requiring deep reasoning and structural precision. Google also provided the Gemini API developer guide and entry points for using it in Google Antigravity and Android Studio. On the same day, Google disclosed an unattended experiment: letting Argon optimize the open-source framework google/adk-python, which has about 205,000 lines of code, within 11 hours. The model proposed 6 optimizations; after human review, 2 were merged into the main branch, memory session preparation time for long sessions was reduced by about 40%, and cold-start import speed improved by 21%. On October 14, Sundar Pichai called this a good summary of the safety and alignment work on Argon 4; Ethan Mollick commented that Gemini 4 is a major leap compared with previous Gemini models but still lags on complex multi-agent tasks. On October 15, Google began rolling out Gemini 4 Argon support to the AI programming tool Antigravity.

English translation is being prepared.

English translation is being prepared.

Elorian releases its first AI model, Glimpse

On October 17, 2026, Elorian released its first AI model, Glimpse. According to Testing Catalog, Glimpse can, during its thinking process, return to a specific point in a video, zoom in on details, or replay half a second of video at a higher frame rate before answering. Elorian's own testing shows that Glimpse outperforms GPT-6 Astra in reading scientific charts and describing short videos; in a simulated task of planning for a robotic arm to place cream cheese into a bowl, Glimpse cost $0.035, compared with $0.249 for GPT-6 Astra; in a demonstration, Glimpse solved 5 mazes of increasing difficulty, while Gemini 3.8 Flash solved 1/5. Another report the same day said that Elorian positions Glimpse as the first visual reasoning model, used to compare scene changes at different points in a video, and calls this frame-based reasoning visual thinking; during its reasoning process it can return to the source video, crop regions, extract specified frames, or run code to measure and compare what it sees, a mechanism called visual scratch space; co-founder Andrew M. Dai said it has frontier performance competitive with Astra on key public visual reasoning benchmarks. At present, there have been no further evidence-supported developments since the model's release.