Search “Search”

English translation is being prepared.

MiniMax M3.1 Flash Tops Search Efficiency Leaderboard

Parallel Web Systems' search capability efficiency leaderboard shows MiniMax M3.1 Flash in first place, with web research performance comparable to DeepSeek V4.1 Flash (1017) at roughly 3x lower cost. MiniMax officially reposted and thanked them, emphasizing that the model is very fast.

English translation is being prepared.

From Afghanistan to Dark Web Search: Microsoft Research Catalyst Lab head Chris White on how technology can have real-world impact

In a podcast, Microsoft Research Catalyst Lab head Chris White reflected on his journey from a Harvard postdoc to a DARPA program manager, then to head of advanced technology in the Afghanistan war zone, and later to dark web search tools to combat human trafficking. He received the 2016 Presidential Award for Extraordinary Efforts to Combat Trafficking in Persons for this work, and emphasized the critical role of problem selection in innovation.

English translation is being prepared.

One Search, Two Layers of Optimization: PyTorch Helion Introduces Compiler Controls with NVIDIA CompileIQ

PyTorch Helion can now search ptxas compiler control parameters through NVIDIA CompileIQ, jointly with the existing kernel configuration search, further tapping performance headroom on the roughly 27 kernels that have already been pre-tuned. CompileIQ packages tuning results as a portable Advanced Control File (ACF); Helion supports specifying a fixed ACF, and also supports searching multiple ACFs together as one dimension of autotuning. These compiler controls are exposed by ptxas starting from CUDA 13.3.

English translation is being prepared.

DeepSeek Opens Up Search and Fetch Search APIs, Charging $0.75 per Thousand Requests

DeepSeek is opening its in-house search engine to external developers through two new APIs, Search and Fetch, which can retrieve real-time web search results and webpage body text. It charges $0.75 per 1,000 successful requests, requires no additional model calls, and can be used in applications powered by other vendors' models. Search returns ranked results with titles, URLs, snippets, and summaries, while Fetch returns the content of a single webpage in Markdown format; the company says the engine handles hundreds of millions of requests per day and its index contains hundreds of billions of webpages. The release coincides with a minor version update, DeepSeek-V4.1-Flash-1017. The API documentation shows that each account is limited to 5 Search and 5 Fetch requests per second, a single search returns at most 50 results, and only requests that return results are billed.

English translation is being prepared.

TinyFish Makes Search Free: Breaking Down the Real Cost of AI Agents Reading Web Pages

TinyFish announced that, within limits, it will waive retrieval fees for Search and Fetch: Search is limited to 30 requests per minute and 500 per hour, while Fetch is limited to 150 URLs per minute and 1,000 per day. The article 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 can 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 about 59 seconds per task.

English translation is being prepared.

Meta Accused of Failing to Comply with Brazilian Court Order; Meta AI and Search Ranking Questioned Again

Sleeping Giants Brasil and CTRL+Z have submitted new evidence to Brazil's Superior Electoral Court, alleging that tools on Meta's Instagram and WhatsApp still favor Senator Flávio Bolsonaro and associate President Lula with derogatory content, even though the court has ordered Meta to ensure both candidates receive equal treatment. The evidence concerns GIF search, hashtag ranking, and answers from the Meta AI built into WhatsApp, and the organizations are demanding that Meta make corrections within 24 hours and disclose the rules for ranking and AI-generated answers. Judge Bellizze had previously set a daily fine of 500,000 reais (about $99,000) and warned that it could rise to 1 million reais (about $198,000) if noncompliance continues; Meta denies intentionally favoring any candidate and has attributed earlier problems to technical glitches.

English translation is being prepared.

DeepSeek V4.1-Flash updated to V4.1-Flash-1017 and opens search and fetch APIs

DeepSeek V4.1-Flash has been updated to version V4.1-Flash-1017; the official web, App and API have been updated in sync, and the model weights are available for download on HuggingFace and ModelScope. In the DeepSWE v1.1 test, the new model's max tier achieved a single-task pass rate of 77.0%, 2.8 percentage points higher than the previous version, and the average cost per task dropped from about 3.3 yuan to about 1.3 yuan. DeepSeek also opened to the public the Search and Fetch APIs of its self-developed search engine; both interfaces are priced at 5 yuan per thousand calls, and this part of access via server-side tool calls will be charged per call starting from 00:00 Beijing time on November 1.

English translation is being prepared.

Meta's quarterly ad revenue gap with Google Search narrows to about $4 billion

Meta's advertising business is narrowing its revenue gap with Google Search, with quarterly revenue now differing by only about $4 billion. AI has increased Meta's ad delivery volume by 14% and raised prices by 12%, because it can better predict users' needs before they search.

Imperial College London paper: search-augmented LLMs often cite web pages they never opened

Three authors from Imperial College London have published the paper "Retrieved but Not Read: Understanding Why Search-Augmented LLMs Fail to Use Available Evidence," examining why search-augmented LLMs frequently cite web pages they never actually open. The study tested GPT-5.6 Luna and six open-weight configurations including GLM-5.3-Flash, gpt-oss-120b, gpt-oss-20b, gemma-4-31b, and qwen3-8b using 225 facts that changed after February 16, 2026, finding that models often answer based solely on search snippets without opening the source pages; under the same set of search results, the proportion of cases where no page was opened ranged from 2.6% to 76%, and when Luna used its built-in search, 53.6% of turns relied only on snippets. When answering from snippets alone, outdated answers accounted for 6.4% and misleading citations for 6.4%, whereas when pages were opened and cited, these fell to 1.3% and 2.3% respectively; placing the correct page directly into the prompt significantly reduced errors, but some models still favored old information even when shown the new facts.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.