All news

SShao Meng12:53Analysis

AI score59

DeepLearning.AI Launches Context Engineering Course, Teaching Agent Context Management

DeepLearning.AI has launched a new Context Engineering course, taught by Chroma CEO Jeff Huber and CMPND CEO Drew Breunig, framing context engineering as the main thread for keeping Agents reliable and token-efficient in complex multi-step tasks. The course builds a harness from scratch in Python and uses traces and evals to quantify each change, covering three capabilities—Agentic Search, Context Management, and Agentic Memory—across four modules, with labs mostly consisting of about 1-hour coding exercises. The course assumes intermediate Python, familiarity with LLM applications and tool calling, and basic retrieval.

#Tutorials / practice#Agents#RAG

PPromptArmor12:53First-partyAnalysis

AI score60

PromptArmor testing found that DeepSeek, Kimi, GLM, and Qwen are 88% more susceptible to cyberattacks than…

PromptArmor tested its most widely circulated attack samples across multiple models and found that DeepSeek, Kimi, GLM, and Qwen models are 88% more susceptible to cyberattacks than GPT, Claude, and Gemini models. The tests covered multiple cyber risk dimensions, including susceptibility to phishing, data exfiltration, and running malware, and the authors described the differences as significant.

#Safety / alignment#Evaluations / benchmarks#DeepSeek

PPromptArmor12:53First-partyAnalysis

AI score52

PromptArmor tested multiple models for cyberattack risk, claiming DeepSeek, Kimi, GLM, and Qwen are more...

PromptArmor said it used multiple highly viral attack samples to test cyber risks including phishing, data exfiltration, and propensity to run malware, and that DeepSeek, Kimi, GLM, and Qwen models were 88% more likely to be attacked than GPT, Claude, and Gemini. This conclusion comes from its main tweet; the specific testing methodology and sample size were not explained in the tweet.

#Safety / alignment#Evaluations / benchmarks#DeepSeek

SShao Meng12:53Analysis

AI score71

Addy on how to use Claude Code in the Claude Opus 5.5 era

Addy Osmani discusses how to use Claude Code in the Claude Opus 5.5 era, arguing that starting with the Claude 5 series, users should reduce piling up prompts and hand-holding task breakdowns, and instead use a single prompt that clearly states the goal and acceptance criteria, letting the model run to completion on its own and then stop. He recommends starting from /effort medium; when the goal is unclear, first have Claude interview you to produce a spec, then open a new session to implement it. For Claude Opus 5 and Claude Fable 5, he removed more than 80% of the Claude Code system prompt, with no noticeable drop in coding benchmarks. He also mentioned that the Claude Code interactive terminal recently defaulted to auto mode, and said this is not a safety guarantee; sub-agents should be split less by hand, and a single agent plus skills often yields better quality.

Quoted postAddy Osmani · @addyosmani

https://t.co/1txsLAxxM5

Same story · 2 reports

#English text is being prepared.#Agents#English text is being prepared.

SShao Meng12:53Analysis

AI score58

Matt Palmer demonstrates using Cursor Cloud Agent to generate a UI motion video against real Slack

Matt Palmer demonstrated opening Chrome and real Slack on the Linux remote desktop of a Cursor Cloud Agent, having the Agent generate a Slack UI motion video with code, and then iterating and revising after checking easing and motion patterns against the real interface. He broke the process down into an inner loop (the Agent does the work, checks against the real thing, and fixes) and an outer loop (a human provides references, preferences, and permissions and accepts the output), and proposed the Do→Codify→Generalize ladder, packaging one success into a Skill and then generalizing it. He emphasized that Slack was not decompiled; he merely gave the Agent a reference and asked it to extract visual patterns by comparison.

Quoted postmatt palmer · @mattyp

Agents rebuilt Slack as video (using code) by looking at the real thing 0:00 Why 0:25 @cursor_ai agents & computer use 1:03 Video as code 1:56 Slack 2:54 A chat, an agent, a computer 3:53 Inner loop and outer loop 5:04 The ladder 7:47 Skills 8:14 Feedback loops 10:02 Recap https://t.co/S91yI9vLop

#Tutorials / practice#Agents#English text is being prepared.

RRohan Paul12:52Analysis

AI score71

Gemini 4 Carbon Tops the Artificial Analysis Intelligence Index with 59 Points

Google's Gemini 4 Carbon ranks first on the Artificial Analysis Intelligence Index with 59 points, 1 point higher than Claude Opus 5.5's 58 points and 2 points higher than GPT-6 Astra's 57 points. Artificial Analysis found that Carbon won 4 of 9 benchmarks (HLE-Diamond, Terminal-Bench-Science, AA-Omniscience, AutomationBench-AA), but it does not lead in coding and knowledge work. Calculated at standard prices, it costs $7.76 per index task, about 23% lower than Opus 5.5's $10.04, but higher than GPT-6 Astra's $6.25.

Quoted postDemis Hassabis · @demishassabis

No.1 💎 https://t.co/k4Werb6ZGV

#Evaluations / benchmarks#Google#Anthropic

AArtificial Analysis12:52Analysis

AI score71

Artificial Analysis tests GitHub Copilot Fusion: $5.59 cost per task…

In Coding Agent Index v2.0, Artificial Analysis benchmarked GitHub Copilot's Copilot Fusion, which costs $5.59 per task, compared with $11.42 for Claude Code with Opus 5.5 (max), while performing only 2 points lower. Copilot Fusion is a runtime orchestration system that selects and coordinates among Anthropic and OpenAI models such as Claude Opus 5.5, Claude Sonnet 5.5, GPT-6 Astra, and GPT-6 Sol. It leads on Terminal-Bench 4.0 with a score of 71, ranks third on FrontierCode with 51 points, and uses 7.2M tokens per task, about half that of Opus 5.5 in Claude Code.

Same story · 6 reports

#Product updates#Agents#English text is being prepared.

AArtificial Analysis12:52Analysis

AI score59

Artificial Analysis releases the image-to-video benchmark AA-Video-I2V v2.1

Artificial Analysis launches the image-to-video evaluation benchmarks AA-Video-I2V v2.1 and AA-Video-I2V-Silent v2.1, ranking models at up to 1080p resolution across 10 use-case categories, 10 capabilities, and multiple visual styles. The two versions are based on more than 58,000 human preference votes across over 1,000 prompts and more than 44,000 votes across 500 prompts, respectively, with only 5% of prompts made public in the public arena and rotated monthly. In the first version of the leaderboard, Dreamina Seedance 2.5 ranks first overall, statistically tied with the Seedance 2.0 and MiniMax H3 series, at a price of $34.12/min; the open-weight MiniMax H3 (768p) ranks third, at a price of $4.80/min.

#Evaluations / benchmarks#Video#Image generation

AArtificial Analysis12:52Analysis

AI score80

Gemini 4 Carbon Tops the Artificial Analysis Intelligence Index…

Artificial Analysis evaluations show that Google's Gemini 4 Carbon tops Intelligence Index v5 with 59 points, ranking first in four of the 9 evaluations: HLE-Diamond, Terminal-Bench-Science, AA-Omniscience, and AutomationBench-AA. On the Coding Agent Index, Gemini 4 Carbon (high) paired with Antigravity CLI scores 59 points, 3 points behind Claude Code paired with Claude Opus 5.5 (max, 62 points). The model is currently available only to some users and has not yet been publicly released; it is priced at $4/$20 per 1M input/output tokens with a 50% discount, has a 1M-token context window, and supports text, image, video, and voice input.

Same story · 34 reports

#Evaluations / benchmarks#Reasoning#English text is being prepared.

AArena12:52Analysis

AI score70

Gemini 4 Carbon ranks 2nd on Agent Arena with a net improvement of 16.2%

Arena.ai says Gemini 4 Carbon ranks 2nd in the real-task-based Agent Arena with a net improvement of 16.2%, and sits on the Pareto frontier for cost, time, and output tokens per task. Compared with Gemini 4 Argon, which ranks 7th with an 11% improvement, Carbon is 5 places higher and 5.2 percentage points better; the gains mainly come from a +13.2 percentage point increase in the positive-feedback-to-complaint ratio, +5.6 percentage points in confirmed success, and +4.8 percentage points in steerability. Arena says Carbon's availability will be announced by Google DeepMind.

Quoted postArena.ai · @arena

Exciting news: Gemini 4 Carbon (High) from @GoogleDeepMind just made a huge debut on Arena, landing #1 in Text Arena, #2 in Agent Arena, and #4 in Code Arena: WebDev. Back at the top, Gemini 4 Carbon delivers frontier intelligence and reshapes the Pareto frontier across three arenas. In Agent Arena, Gemini 4 Carbon ranks #2 with a +16.2% net improvement across +21.9K real-world agentic sessions from our global community, surpassing top models GPT-6 Astra and Fable-5.1. At $1.80 median cost per task, Gemini 4 Carbon (High) costs 72% less than the #1 ranked Claude Opus 5.5 (Max). Its standouts by key signals: - #1 Steerability (+18.6%): can the model course-correct when you push back? - #2 Praise vs Complaint (+37.9%): implicit sentiment in users' reactions - #3 Confirmed Success (+18.7%): explicit “yes, that worked” from users In Code Arena: WebDev, Carbon scored 1768 pts landing #4: - Just 2 pts below #3 Claude Sonnet 5.5 (xHigh), at 1770 pts - 89 pts above #11 Gemini 4 Argon (High), at 1679 pts By category, it ranks #3 in Gaming and Simulations, and #5 in Brand & Marketing. In Text Arena, Gemini 4 Carbon (High) is the new #1 with 1528 pts, now ahead of Gemini 4 Argon (High) at #2 with 1520 pts. Congrats to the @GoogleDeepMind team on this huge milestone!

Same story · 2 reports

#Evaluations / benchmarks#Agents#Google

AArena12:52Analysis

AI score65

GPT-6.1 Terra ranks 8th in Agent Arena, with a median cost of $0.62 per task

Arena's empirical results, based on nearly 9,000 agent sessions, show that OpenAI's GPT-6.1 Terra (Max) ranks 8th in Agent Arena, with a net improvement of +10.6% and a median cost of $0.62 per task. This cost is 43% lower than the $1.09 of 7th-place Gemini 4 Argon (High), 30% lower than the $0.89 of 6th-place GPT-6.1 Sol (Max), and 88% lower than the $5.16 of 5th-place GPT-6 Astra (Max). In the subcategories, it ranks 3rd in Bash Reliability (+13.2%), 7th in Steerability (+13.3%), and 9th in Praise vs Complaint (+16.8%).

Quoted postArena.ai · @arena

Real-world results are in: GPT-6.1 Terra (Max) by @OpenAI just landed #14 in the Code Arena: WebDev with 1671 pts, delivering Pareto level results! That’s +151 pts over GPT-5.6 Terra, now at #56. At a median cost of $0.40 per task, GPT-6.1 Terra costs 28% less than GPT-5.6 Terra ($0.56 per task), reshaping the Pareto frontier. Compared to other models in the family, GPT-6.1 Terra (Max) is incredibly efficient: - 81 pts below #4 GPT-6.1 Sol (1752 pts), at 75% lower median cost per task. - Just 6 pts below #11 GPT-6 Sol (1677 pts), at 76% lower median cost per task Congrats to the @OpenAI team on this release!

#Evaluations / benchmarks#Agents#OpenAI

AArena12:52Analysis

AI score60

Arena Analysis: Agent Arena Cost per Task: GPT-6.1 Sol at $1.91

Arena's statistics on Agent Arena show that OpenAI's GPT-6.1 Sol (Max) costs $1.91 per task, 85% less than Claude Opus 5.5 (Max) at $12.51 and 87% less than GPT-6 Astra (Max) at $14.93. Arena notes that the sources of these two cost advantages differ: relative to Astra, Sol's advantage comes mainly from a lower token unit price; relative to Opus, it comes mainly from consuming fewer tokens per task. Cost per task is calculated by multiplying the number of input, output, and cache-read tokens by their respective unit prices, and Arena pinpoints the causes of cost reductions by comparing prices and token usage.

#Evaluations / benchmarks#Agents#OpenAI

AArtificial Analysis12:52Analysis

AI score60

Fish Audio Drama 3 ranks first among open-weight models in two speech arenas, and first in pronunciation robustness

Artificial Analysis evaluations show that Drama 3, released by Fish Audio, has an Elo of 1,278 in the Provider Voice scenario of Text to Speech (1,512 arena appearances), ranking 6th overall and 1st among open-weight models; in the Controlled Voice scenario, its Elo is 1,085 (1,958 appearances), ranking 12th overall and 1st among open-weight models. On pronunciation robustness, Drama 3 ranks 1st with 94.6% (2,157 rated clips), higher than the 93.2% of Ming-3.0-tts and Drama 3 Preview. Its generation speed is 66.6 characters per second, its pricing is $15/1M characters, and its weights are open under the Fish Audio Community License.

#Evaluations / benchmarks#Speech#Open-source ecosystem

AArtificial Analysis12:52Analysis

AI score62

Phonic 1 Tops the Artificial Analysis Speech-to-Speech Index, Leading Grok with 87.1 Points…

Artificial Analysis evaluations show that Phonic's Phonic 1 (High) ranks first among 22 models on the Speech to Speech Index with 87.1 points, leading Grok Voice 3.0's 86.9 points by a narrow margin. The model ranks first on the Tau Voice agent task with 79.9%, leading the second place by 5.4 percentage points; it ranks third on Big Bench Audio with 98.9%, and its Time to First Audio is 0.61 seconds. Phonic 1 is available through the Phonic API, with input audio costing $0.07 per minute and $4.20 per hour.

Same story · 2 reports

#Evaluations / benchmarks#Speech#Agents

AAlibaba Cloud Developer Community AI12:51Analysis

AI score60

How Hermes Writes Agent Learning into Files: A Three-Layer Design of Memory, Skills, and Knowledge Base

The Hermes Agent open-sourced by Nous Research externalizes learning into files, using MEMORY.md and USER.md to store memory, SKILL.md to store reusable skills, and a wiki to store domain knowledge, while the underlying model weights remain unchanged. The article breaks down its closed-loop learning mechanism, including post-task background review that automatically creates and improves skills. The built-in llm-wiki skill 2.1.0 maintains the knowledge base with a three-layer structure of raw/, wiki, and SCHEMA.md, along with disciplines such as page-creation thresholds, bidirectional links, source-text hashing, and contradiction registration, and comes with an inspection mechanism.

#Tutorials / practice#Agents#Open-source ecosystem

Hheise online12:51Analysis

AI score58

Artificial Analysis releases Intelligence Index v5, with 60% of tasks now undisclosed

Independent benchmarking organization Artificial Analysis has updated its main index to Intelligence Index v5. The index aggregates results from nine benchmarks and, for the first time, uses 60% undisclosed tasks to reduce the impact of models overfitting to public questions. The two newly added tests are Terminal-Bench-Science, which contains 70 agentic science tasks, and FrontierCode 2.0, recently launched by Cognition, which examines whether repository maintainers would adopt an agent's changes in real coding tasks; Humanity's Last Exam Diamond, AA-Omniscience v1.1 and CritPt v2 were upgraded, while GDPval-AA v2.1, SciCode and AA-LCR v1.1 were removed. Rankings changed after the test adjustments: Claude Opus 5.5 briefly led before being overtaken by the recently released Gemini 4 Carbon, Grok 4.7 and Kimi K3.1 Pro saw their scores decline, and the gap between the strongest open-source model and the top spot widened from 3 points to 12 points. The original article also notes that undisclosed benchmarks cannot be externally reviewed, making potential errors hard to detect.

Same story · 8 reports

#Evaluations / benchmarks#English text is being prepared.#Agents

GGoogle Developers Blog12:48First-partyAnalysis

AI score70

Google A2UI Team and CopilotKit Explain How to Build Generative UI Applications

The Google A2UI team and CopilotKit CEO Atai Barkai explain how to build generative UI applications using the A2UI open standard and CopilotKit. A2UI describes interfaces with declarative JSON, and the client only renders components from a trusted component catalog; the same response can be rendered to React, Angular, SwiftUI, Jetpack Compose, and Flutter. The article provides a code example for integrating CopilotKit with Google ADK, and explains the bidirectional flow, implemented via the AG-UI protocol, in which user actions are sent back and the agent regenerates the interface.

#Tutorials / practice#Agents#English text is being prepared.

OOfficial Pydantic AI article12:47First-partyAnalysis

AI score68

Pydantic compares code mode and tool calling across 5,276 runs

Michael Pfaffenberger of Pydantic used 5 small Pydantic AI applications and 8 models to conduct 5,276 scored runs, comparing code mode with direct tool calling. On long tasks, code mode reduced average cost by 58%, latency by 51%, and increased success rate by 16 percentage points; however, for short tasks with 5 items, 3 out of 5 tasks were instead 21% to 38% more expensive. The author argues that models are better at writing loops than at acting as the loop, and the gap widens as tasks get longer.

#Evaluations / benchmarks#Agents#English text is being prepared.

LLatent Space12:47Analysis

AI score54

Elorian Releases DecoyCount Benchmark: Frontier Models All Score Below 50% on Visual Counting Accuracy

Elorian and Mercor jointly released the DecoyCount benchmark, which tests whether models truly count objects or rely on visual shortcuts. Frontier models all scored below 50% accuracy, with the highest being GPT-6 Astra at 43.17%. Elorian says its visual reasoning model Glimpse scored 50.48% in a single run on 1,660 questions, making it the only model to exceed 50%, but this result comes from internal testing and did not use an agent or tool framework, whereas the six models in the paper each took the average of three runs. Elorian CEO Andrew Dai said the benchmark examines both vision and reasoning, and can distinguish whether a model failed to see the objects or miscounted them.

#Evaluations / benchmarks#Multimodal#Reasoning

BBeijing News12:47Analysis

AI score67

Beijing News tests Qwen Go! and Lingguang 2.0: personal agent competition shifts toward autonomous task handling

Beijing News Shell Finance reporters tested Alibaba's Qwen Go! and Ant's Lingguang 2.0. The former set up a daily 8 a.m. medication reminder with a single sentence and checked it item by item, while the latter, after authorization on Alipay, sorted out the past six months of bills. The article also reviews the layouts of Meta Muse, OpenAI dots, as well as ByteDance and Tencent, and points out that cross-ecosystem permissions are currently the hardest wall to break down, with ByteDance's SAEP published rejection list reaching 155 apps.

#English text is being prepared.#Agents#OpenAI