←Search results

Search “Reasoning”

By relevance · Total 3279 results Clear search

Friday

  1. AIhub

    AIhub October 2026 Monthly Roundup: Neuro-Symbolic Planning, Time Series Forecasting, and Home Robots

    AIhub has released its October 2026 monthly roundup, summarizing recent content on neuro-symbolic planning, clinical time series forecasting, home robots, and more. This issue includes an interview with AI pioneer Michael Littman, as well as an introduction by Raphael Memmesheimer, five-time RoboCup@Home champion, to work on semi-supervised object segmentation annotation, which won the RoboCup 2026 Best Paper Award. The roundup also features an interview from ICML 2026 on research showing that vision-language models cannot reason about physical transformations.

    AIhub has released its October 2026 monthly roundup, summarizing recent content on neuro-symbolic planning, clinical time series forecasting, home robots, and more. This issue includes an interview with AI pioneer Michael Littman, as well as an introduction by Raphael Memmesheimer, five-time RoboCup@Home champion, to work on semi-supervised object segmentation annotation, which won the RoboCup 2026 Best Paper Award. The roundup also features an interview from ICML 2026 on research showing that v

    Matched fieldsSummary

    #ai-vertical#Industry news
  2. Yicai

    Hainan Computing Power and Tencent Cloud Reach Strategic Cooperation to Jointly Build an Integrated "AI + Computing Power" Foundation for the Free Trade Port

    Hainan Province Computing Power Technology Co., Ltd. and Tencent Cloud signed a strategic cooperation agreement to jointly create an integrated "AI + computing power" solution for the Hainan government and enterprise market. Tencent Cloud will provide resources such as high-performance computing, intelligent computing power, storage, and networking to support the large-scale training and inference needs of the provincial-level artificial intelligence public service platform. The two parties will also carry out business cooperation around large model invocation tokens, and, leveraging the institutional advantages of the free trade port's cross-border data flow regime, explore commercialization paths such as token going global and cross-border computing power export.

    Hainan Province Computing Power Technology Co., Ltd. and Tencent Cloud signed a strategic cooperation agreement to jointly create an integrated "AI + computing power" solution for the Hainan government and enterprise market. Tencent Cloud will provide resources such as high-performance computing, intelligent computing power, storage, and networking to support the large-scale training and inference needs of the provincial-level artificial intelligence public service platform. The two parties will

    Matched fieldsSummary、Body

    #chinese#Industry news#English text is being prepared.#English text is being prepared.
  3. Phoenix Technology

    XHEART Releases Large-Model Chip X7 Ultra with 700 TOPS of Compute

    XHEART has released the X7 Ultra, a new-generation chip designed for billion-parameter-level smart-driving large models, with 700 TOPS of compute. The company says that in real-world tests using a partner's production Transformer smart-driving model, compute utilization reached 98.7%, and that while matching the model performance of industry flagship chips, memory bandwidth consumption dropped by 80% and inference energy efficiency improved by 100%. The X7 series has shipped a cumulative 200,000 units in 8 months of mass production, equipping 10 production vehicle models, with the first launch model being the SAIC Volkswagen ID.ERA 9X; at the same event, Momenta released the R7 Ultra world model, whose parameter count is 5 to 10 times higher than the previous generation.

    XHEART has released the X7 Ultra, a new-generation chip designed for billion-parameter-level smart-driving large models, with 700 TOPS of compute. The company says that in real-world tests using a partner's production Transformer smart-driving model, compute utilization reached 98.7%, and that while matching the model performance of industry flagship chips, memory bandwidth consumption dropped by 80% and inference energy efficiency improved by 100%. The X7 series has shipped a cumulative 200,000

    Matched fieldsSummary、Body

    #chinese#Product updates#English text is being prepared.#Reasoning
  4. Securities Times

    Hainan Computing Power and Tencent Cloud Reach Strategic Cooperation to Build an Integrated "AI + Computing Power" Foundation for the Free Trade Port

    Hainan Provincial Computing Power Technology Co., Ltd. and Tencent Cloud signed a strategic cooperation agreement to jointly build an integrated "AI + Computing Power" solution for Hainan's government and enterprise market, centered on computing power infrastructure, large model Token services, and AI office applications. Tencent Cloud will provide Hainan Computing Power and its customers with high-performance computing, intelligent computing power, storage, and network resources to support the large-scale training and inference needs of the provincial artificial intelligence public service platform. The two parties will also explore commercialization paths such as Token going global and cross-border computing power export.

    Hainan Provincial Computing Power Technology Co., Ltd. and Tencent Cloud signed a strategic cooperation agreement to jointly build an integrated "AI + Computing Power" solution for Hainan's government and enterprise market, centered on computing power infrastructure, large model Token services, and AI office applications. Tencent Cloud will provide Hainan Computing Power and its customers with high-performance computing, intelligent computing power, storage, and network resources to support the

    Matched fieldsSummary、Body

    #chinese#Industry news#English text is being prepared.#English text is being prepared.
  5. Rohan Paul

    Anthropic adds post-response self-assessment to Claude to flag potential misuse in conversations

    Anthropic has trained Claude to perform a preliminary safety assessment of the conversation after each response; this technique, called self-assessment, has since it was enabled in June surfaced misuse that other safeguards were not designed to catch. The mechanism takes advantage of Claude having already read the conversation, costs less than classifiers, and improves in accuracy as the model improves. Rohan Paul adds that Anthropic's most accurate misuse detector is the Claude Mythos Preview classifier, which must first read all safety policies and is too expensive to run; Anthropic therefore trained Claude to replicate its judgments without reading the policies or the reasoning process, and on Fable 5 traffic this replica matched 78% of the detector's flags, rising to 94% for high-priority cases, with a false positive rate of 1%.

    Anthropic has trained Claude to perform a preliminary safety assessment of the conversation after each response; this technique, called self-assessment, has since it was enabled in June surfaced misuse that other safeguards were not designed to catch. The mechanism takes advantage of Claude having already read the conversation, costs less than classifiers, and improves in accuracy as the model improves. Rohan Paul adds that Anthropic's most accurate misuse detector is the Claude Mythos Preview c

    Matched fieldsSummary、Body

    #ai-news#Safety / alignment#Anthropic#Product updates
  6. Tencent HunyuanSelected

    Tencent Hunyuan Releases Hy-Nomas Technical Report and Open-Sources Hy-Nomas-Bench

    Tencent Hunyuan has released the technical report for Hy-Nomas, an interactive video world model, and has open-sourced Hy-Nomas-Bench for reproducible research. Officially, Hy-Nomas was also used to reproduce Genie 4's examples and a comparison video was released, along with an online demo that requires no waitlist. Earlier introductions stated that Hy-Nomas supports free interaction with text prompts and camera trajectories, long-horizon generation with world memory, lightweight LoRA adaptation for fields such as robotics, and optimization for inference on consumer-grade GPUs.

    Tencent Hunyuan has released the technical report for Hy-Nomas, an interactive video world model, and has open-sourced Hy-Nomas-Bench for reproducible research. Officially, Hy-Nomas was also used to reproduce Genie 4's examples and a comparison video was released, along with an online demo that requires no waitlist. Earlier introductions stated that Hy-Nomas supports free interaction with text prompts and camera trajectories, long-horizon generation with world memory, lightweight LoRA adaptation

    Matched fieldsSummary

    #company#Model releases#Video#Open-source ecosystem

Thursday

  1. ARC Prize

    ARC Prize: Strong correlation among the three ARC-AGI benchmark scores

    After plotting ARC-AGI model scores, ARC Prize found a strong association among the three benchmarks: the power curve v2 = v1^4.53 can explain about 98% of the variation in ARC-AGI-2 scores, and models that are on par on ARC-AGI-1 may still differ by more than 20 percentage points on ARC-AGI-2. A similar trend appears on ARC-AGI-3, but the low-reasoning tier of Gemini 4 Argon is a clear exception: only 42.1% on ARC-AGI-2, yet 97.5% on ARC-AGI-3.

    Matched fieldsSummary

    #company#Evaluations / benchmarks#Reasoning
  3. ARC Prize

    ARC Prize: The three generations of ARC-AGI benchmarks show a natural progression

    ARC Prize stated that the relationship among ARC-AGI-1, ARC-AGI-2, and ARC-AGI-3 shows a natural progression, and that future benchmarks will continue to help measure progress and distinguish increasingly capable models. The thread also mentioned that there is a similar relationship between the scores of ARC-AGI-2 and ARC-AGI-3, with Gemini 4 Argon under low-reasoning settings being a clear exception: ARC-AGI-2 only 42.1%, while ARC-AGI-3 reached 97.5%.

    Matched fieldsSummary

    #company#Evaluations / benchmarks#Reasoning#English text is being prepared.
  4. ARC PrizeSelected

    ARC Prize publishes Meta Muse Spark Open's evaluation results on ARC-AGI

    ARC Prize published the validation results of Meta's Muse Spark Open on ARC-AGI: ARC-AGI-1 at 89.0% with $0.29 per task, ARC-AGI-2 at 57.5% with $0.73 per task, and ARC-AGI-3 at 0.1% under the standard harness with $3.9K. ARC Prize states that its v1/2 scores are close to those of Haiku 5.5 and GPT-6 Luna, but the cost is about 9 times that of the latter; the model is priced on OpenRouter at $1.25 per million input tokens and $4.25 per million output tokens. As the reasoning tier increases, ARC-AGI-2 rises from 29.4% at medium (averaging 55k tokens/task) to 57.2% at xhigh (180k tokens/task).

    ARC Prize published the validation results of Meta's Muse Spark Open on ARC-AGI: ARC-AGI-1 at 89.0% with $0.29 per task, ARC-AGI-2 at 57.5% with $0.73 per task, and ARC-AGI-3 at 0.1% under the standard harness with $3.9K. ARC Prize states that its v1/2 scores are close to those of Haiku 5.5 and GPT-6 Luna, but the cost is about 9 times that of the latter; the model is priced on OpenRouter at $1.25 per million input tokens and $4.25 per million output tokens. As the reasoning tier increases, ARC-

    Matched fieldsSummary

    #company#Evaluations / benchmarks#Reasoning#Meta
  5. ARC PrizeSelected

    ARC Prize tests Meta Spark Open: ARC-AGI-2 score 57.5%, cost about 9 times that of models with the same score

    ARC Prize tested Meta Spark Open on ARC-AGI: it scored 89.0% on ARC-AGI-1, 57.5% on ARC-AGI-2, and only 0.4% on ARC-AGI-3. Its v1/v2 scores are close to those of Haiku 5.5 and GPT-6 Luna, but its cost is about 9 times theirs. Across providers on OpenRouter, the model costs $1.25/M for input and $4.25/M for output, while Haiku 5.5 and GPT-6 Luna both cost $0.10/M for input and $0.50/M for output. As the reasoning tier increases, ARC-AGI-2 rises from 29.4% at medium (averaging 55k tokens/task) to 57.2% at xhigh (180k tokens/task).

    ARC Prize tested Meta Spark Open on ARC-AGI: it scored 89.0% on ARC-AGI-1, 57.5% on ARC-AGI-2, and only 0.4% on ARC-AGI-3. Its v1/v2 scores are close to those of Haiku 5.5 and GPT-6 Luna, but its cost is about 9 times theirs. Across providers on OpenRouter, the model costs $1.25/M for input and $4.25/M for output, while Haiku 5.5 and GPT-6 Luna both cost $0.10/M for input and $0.50/M for output. As the reasoning tier increases, ARC-AGI-2 rises from 29.4% at medium (averaging 55k tokens/task) to

    Matched fieldsSummary

    #company#Evaluations / benchmarks#Reasoning#OpenAI
  6. ARC Prize

    ARC Prize announces Meta Muse Spark Open's evaluation results on various versions of ARC-AGI

    ARC Prize announced the validation results of Meta Muse Spark Open on ARC-AGI: ARC-AGI-1 is 89.0% at $0.29 per task, ARC-AGI-2 is 57.5% at $0.73 per task, and ARC-AGI-3 is 0.4% at $3.0K under the provider adapter harness and 0.1% at $3.9K under the standard harness. ARC Prize states that its scores on v1/v2 are similar to those of Haiku 5.5 and GPT-6 Luna, but the cost is about 9 times that of the latter; the model is priced on OpenRouter at $1.25 per million input tokens and $4.25 per million output tokens, while Haiku 5.5 and GPT-6 Luna are both $0.10 per million input tokens and $0.50 per million output tokens. On public tasks, as the reasoning scale increases, the ARC-AGI-2 score rises from an average of 55k tokens/task corresponding to 29.4% at the medium tier, to 150k corresponding to 51.0% at the high tier, and then to 180k corresponding to 57.2% at the xhigh tier; ARC-AGI-1 has already reached 92.8% at the high tier and rises to 94.0% at the xhigh tier.

    ARC Prize announced the validation results of Meta Muse Spark Open on ARC-AGI: ARC-AGI-1 is 89.0% at $0.29 per task, ARC-AGI-2 is 57.5% at $0.73 per task, and ARC-AGI-3 is 0.4% at $3.0K under the provider adapter harness and 0.1% at $3.9K under the standard harness. ARC Prize states that its scores on v1/v2 are similar to those of Haiku 5.5 and GPT-6 Luna, but the cost is about 9 times that of the latter; the model is priced on OpenRouter at $1.25 per million input tokens and $4.25 per million o

    Matched fieldsSummary

    #company#Evaluations / benchmarks#Reasoning#Open-source ecosystem
  8. Rohan Paul

    Huang Renxun says doubling model speed can double AI lab revenue

    Huang Renxun stated that, under the premise that demand exceeds capacity, doubling model speed (for example, through software or kernel optimization) can double AI lab revenue. Inference revenue equals the number of tokens served multiplied by the price per token; faster kernels can increase tokens per second under the same GPUs, power, and data center conditions. If all additional tokens can be sold at the current price, a 2x speedup can double revenue with zero additional capital expenditure.

    Matched fieldsSummary、Body

    #ai-news#English text is being prepared.#Reasoning#English text is being prepared.
  9. Testing Catalog

    AI/ML API launches Jev Router, automatically selecting models by difficulty

    AI/ML API has released Jev Router, an MCP connector for agent tools. It uses TypeSafe's Jev to assess the difficulty of each request and then automatically selects the model and tools. Simple requests are routed to lightweight models such as GPT-6 Luna, while high-difficulty reasoning requests go to Claude Opus 5.5, avoiding paying flagship-model rates for simple tasks. Users only need to provide the task; Jev Router first selects the model and tools such as web search and page extraction, then returns the result.

    AI/ML API has released Jev Router, an MCP connector for agent tools. It uses TypeSafe's Jev to assess the difficulty of each request and then automatically selects the model and tools. Simple requests are routed to lightweight models such as GPT-6 Luna, while high-difficulty reasoning requests go to Claude Opus 5.5, avoiding paying flagship-model rates for simple tasks. Users only need to provide the task; Jev Router first selects the model and tools such as web search and page extraction, then

    Matched fieldsSummary、Body

    #hot-signal#Product updates#English text is being prepared.#Agents
  10. Aditya Agarwal

    South Park Commons Uses Five Dimensions to Identify Early-Stage Founders

    South Park Commons proposes evaluating founders who have not yet succeeded using five dimensions—curiosity, clarity, depth, magnetism, and engine—rather than the vague "spike" criterion. The framework has been applied to projects including Baseten (a 2019 bet on inference), Goodfire (2024, interpretability), Gamma, Profound, Density, and Gradient Robotics. SPC emphasizes observing founders' growth slope through long-term in-person community involvement rather than performance across two or three pitch sessions.

    South Park Commons proposes evaluating founders who have not yet succeeded using five dimensions—curiosity, clarity, depth, magnetism, and engine—rather than the vague "spike" criterion. The framework has been applied to projects including Baseten (a 2019 bet on inference), Goodfire (2024, interpretability), Gamma, Profound, Density, and Gradient Robotics. SPC emphasizes observing founders' growth slope through long-term in-person community involvement rather than performance across two or three

    Matched fieldsSummary、Body

    #ai-news#Industry news
  11. Testing Catalog

    Mistral releases Mistral Large 4 (Le Chonk) weights on Hugging Face

    Mistral released the model weights of the new version of Le Chonk (Mistral Large 4) on Hugging Face, targeting agentic coding and cybersecurity scenarios. Mistral claims that the model ranks first among open-source models on Cognition's FrontierCode benchmark and sets a new benchmark for Pareto efficiency on Artificial Analysis, while also improving inference efficiency and reducing inference costs. The model is also claimed to achieve SOTA in cybersecurity defense and has been tested by vendors such as CrowdStrike, Cloudflare, Rubrik, and Zscaler; users can customize and self-deploy it with Mistral Forge.

    Mistral released the model weights of the new version of Le Chonk (Mistral Large 4) on Hugging Face, targeting agentic coding and cybersecurity scenarios. Mistral claims that the model ranks first among open-source models on Cognition's FrontierCode benchmark and sets a new benchmark for Pareto efficiency on Artificial Analysis, while also improving inference efficiency and reducing inference costs. The model is also claimed to achieve SOTA in cybersecurity defense and has been tested by vendors

    Matched fieldsSummary、Body

    #hot-signal#Model releases#Open-source ecosystem#Reasoning
  12. StepFun

    StepFun and Together Host Step 5 Preview Launch Night

    StepFun and Together Compute jointly hosted the Step 5 Preview Launch Night, where the event discussed long-horizon agents, tool-calling reliability, inference costs, and what AI breakthroughs mean for the next generation of startups. The panelists included Dan Fu (Together Compute), Mar Hershenson (Pear VC), Saoud Rizwan (Cline), and Joey Wang (Harvey).

    Matched fieldsSummary、Body

    #company#Industry news#Agents#English text is being prepared.

Wednesday

  1. Saito Hardcore Hacker

    SemiAnalysis Analyzes Anthropic S-1: 90% Incremental Gross Margin and Half-Trillion Compute Commitment

    The SemiAnalysis podcast analyzes Anthropic's S-1, pointing out its 90% incremental gross margin, 40% incremental EBIT, 85% inference gross margin, and 41% compute gross margin, and that it has been profitable for two consecutive quarters on a non-GAAP basis. On customer net revenue retention, customers who spent $1 last year now spend $9, with the Q1 figure at 500%; in addition, 250 customers pay over $10 million annually and 1,900 pay over $1 million. The episode also breaks down the $500 billion to $600 billion compute commitment, which at $100 billion per gigawatt translates to only about 5 gigawatts, and discusses the discrepancy in reporting that the top five customers account for 25% of revenue and that 48% goes through cloud vendor marketplaces.

    The SemiAnalysis podcast analyzes Anthropic's S-1, pointing out its 90% incremental gross margin, 40% incremental EBIT, 85% inference gross margin, and 41% compute gross margin, and that it has been profitable for two consecutive quarters on a non-GAAP basis. On customer net revenue retention, customers who spent $1 last year now spend $9, with the Q1 figure at 500%; in addition, 250 customers pay over $10 million annually and 1,900 pay over $1 million. The episode also breaks down the $500 bill

    Matched fieldsSummary、Body

    #ai-news#Industry news#OpenAI#Anthropic
  2. Mistral AI · X

    Mistral Launches Public Preview of Responses API for Agent Building

    Mistral has released a public preview of the Responses API for building agents on Mistral. Each response returns the model's answer, reasoning process, and tool calls as independent items, making it easier for applications to process them step by step than with Chat Completions. The API is available for all chat models hosted by Mistral, including Large 4.

    Matched fieldsSummary、Body

    #hot-signal#Product updates#Agents#English text is being prepared.