🚨 AI News | TestingCatalog (@testingcatalog.com)

🚨 AI News | TestingCatalog

@testingcatalog.com

Latest AI News on AI Agents, Model Releases, Tools, Leaks, and Rumors 🗞 https://www.testingcatalog.com/

904 followers 67 following3K posts

🚨 AI News | TestingCatalog

@testingcatalog.com

NVIDIA and Meta expand Muse collaboration for AI agents

testingcatalog.comNVIDIA and Meta expand Muse collaboration for AI agentsNVIDIA and Meta are deepening work on Muse AI agents, tuning models, infrastructure and safety tooling to raise performance and cut costs. NVIDIA will also invest $3 billion as Meta scales compute, networking and oversight.

🚨 AI News | TestingCatalog

@testingcatalog.com

Glasser now lets AI agents buy premium data per API call

testingcatalog.comGlasser now lets AI agents buy premium data per API callGlasser lets AI agents buy premium data per call across 2,000+ paid APIs with one key. It cuts costs for low-volume use, shows pricing upfront, supports local and hosted agents, and returns raw provider results with charges.

🚨 AI News | TestingCatalog

@testingcatalog.com

OpenAI launches GPT-6.1 Terra for precise computer use

testingcatalog.comOpenAI launches GPT-6.1 Terra for precise computer useOpenAI launched GPT-6.1 Terra, a faster, lower-cost model for visual reasoning, computer use, and writing. It outperforms GPT-5.6 Terra, nears GPT-6 Astra on key tests, and is now in ChatGPT and the API.

🚨 AI News | TestingCatalog

@testingcatalog.com

Meta is working on avatar sharing for Muse assistant

testingcatalog.comMeta is working on avatar sharing for Muse assistantMeta is testing avatar sharing for Muse on the web, adding social sharing and link copying for animated avatars with four preview states. The feature is unreleased, and support for multiple avatars or full agent sharing remains unclear.

🚨 AI News | TestingCatalog

@testingcatalog.com

Google AI Studio prepares new Managed Agents hub

testingcatalog.comGoogle AI Studio prepares new Managed Agents hubGoogle is preparing a Managed Agents area in AI Studio to configure, trigger, and monitor autonomous agents. The move aligns with Agent Compute Engine, while billing lag and credit limits remain key considerations for agent workloads.

🚨 AI News | TestingCatalog

@testingcatalog.com

Gemini 4 Carbon early test results are in ahead of public release

testingcatalog.comGemini 4 Carbon early test results are in ahead of public releaseGemini 4 Carbon appears to be in early access outside Google, with testers citing stronger 3D output than Argon but weaker UI work and much slower speeds. Leaks suggest launch may be close, though public claims remain unverified.

🚨 AI News | TestingCatalog

@testingcatalog.com

Kimi K3.1 is prepared for upcoming release after Arena testing

testingcatalog.comKimi K3.1 is prepared for upcoming release after Arena testingSigns point to Moonshot nearing a Kimi K3.1 launch, likely the anonymous Arena model Kibitz. Early Arena tests suggest strong coding and 3D output, fast response, and very low cost, though claims remain unconfirmed pending a model card.

🚨 AI News | TestingCatalog

@testingcatalog.com

Canva launches Craft image model with 2K output and API

testingcatalog.comCanva launches Craft image model with 2K output and APICanva launched Craft, its first image-generation model, in Dream Lab and via API. Built for design tasks, it delivers up to 2K images, multilingual prompts and text, ranked outputs, editable layers, and faster, lower-cost inference.

🚨 AI News | TestingCatalog

@testingcatalog.com

Claude Security gets continuous scans and fix tracking

testingcatalog.comClaude Security gets continuous scans and fix trackingClaude Security now continuously scans connected repositories, tracks findings through fixes and regressions, and uses team context to prioritize risk. In public beta for Claude Enterprise, it also adds API, webhooks, and MCP support.

🚨 AI News | TestingCatalog

@testingcatalog.com

Google Cloud adds OpenAI models in private preview

testingcatalog.comGoogle Cloud adds OpenAI models in private previewGoogle Cloud is adding OpenAI’s GPT-6 Astra, Sol, and Luna to Model Garden, letting enterprises access OpenAI alongside 200+ models with unified billing, governance, and security controls on existing cloud commitments.

🚨 AI News | TestingCatalog

@testingcatalog.com

Recursive releases Flint, 40B model for local coding

testingcatalog.comRecursive releases Flint, 40B model for local codingRecursive unveiled Flint, a 40B MoE local coding model with 2.5B active parameters per token. Built in under four months with Eureka Machine, it delivers top sub-50B benchmark results, fast local inference, and Apache 2.0 open weights.

🚨 AI News | TestingCatalog

@testingcatalog.com

Anthropic opens Claude Duet preview for US Pro and Max users

testingcatalog.comAnthropic opens Claude Duet preview for US Pro and Max usersAnthropic launched Claude Duet in US research preview for Pro and Max users. The standalone agent surfaces priorities, works in the background, and requires approval for messages, payments, signatures, and deletions.

🚨 AI News | TestingCatalog

@testingcatalog.com

Mistral launches Voxtral 3 with a voice agent SDK

testingcatalog.comMistral launches Voxtral 3 with a voice agent SDKMistral launched Voxtral 3 with speech-to-text, text-to-speech, and a real-time voice agent SDK. It supports hosted and private deployment, targets enterprise call workflows, and offers multilingual, low-latency voice processing.

🚨 AI News | TestingCatalog

@testingcatalog.com

Perplexity Computer now runs fully locally on Mac

testingcatalog.comPerplexity Computer now runs fully locally on MacPerplexity’s Mac app now offers fully local Computer on Apple silicon, keeping model, agent, and task execution on-device. It can access private files and native apps with permission, giving users control over data and credit use.

🚨 AI News | TestingCatalog

@testingcatalog.com

OpenAI testing new "Aurora" GPT-Image model on Arena

testingcatalog.comOpenAI testing new "Aurora" GPT-Image model on ArenaOpenAI briefly exposed three new image model references in its API, then removed them. Arena tests suggest they may be GPT-Image-3 variants, with red favoring speed and green showing higher quality.

🚨 AI News | TestingCatalog

@testingcatalog.com

Anthropic prepares to release Claude Fable 5.5 next week

testingcatalog.comAnthropic prepares to release Claude Fable 5.5 next weekLeaks suggest Anthropic may launch Claude Fable 5.5 soon, with traces found in Azure, third-party tools, and partner testing. Multiple new Claude codenames also point to a broader near-term model rollout, though nothing is official.

🚨 AI News | TestingCatalog

@testingcatalog.com

ICYMI: Meta's Muse Spark Open is optimized for new Mac Studio

testingcatalog.comICYMI: Meta's Muse Spark Open is optimized for new Mac StudioMeta released Muse Spark Open, an open-weight multimodal model for coding, agentic work, and multi-step reasoning. It has 816B parameters, a 1M-token context window, runs on two Mac Studio M5 Ultras, and is offered via APIs and downloads.

🚨 AI News | TestingCatalog

@testingcatalog.com

Krea launched Krea 3 for fast and precise image editing

testingcatalog.comKrea launched Krea 3 for fast and precise image editingKrea 3 is a new image-editing model for prompt-based 2K edits in under 8 seconds, with region control, bounding boxes, multi-turn edits, LoRA fine-tuning, 4K support, and API access for design, marketing, and commerce teams.

🚨 AI News | TestingCatalog

@testingcatalog.com

Meta launches Muse Wonder and Wonder Full for AI agents

testingcatalog.comMeta launches Muse Wonder and Wonder Full for AI agentsMeta introduced Muse Wonder and Muse Wonder Full for agentic personal and professional work, highlighting memory, sustained problem-solving, coding, domain expertise, API access, and stronger safety controls for sensitive tasks.

🚨 AI News | TestingCatalog

@testingcatalog.com

Google is preparing Nano Banana 3 Pro upgrade

testingcatalog.comGoogle is preparing Nano Banana 3 Pro upgradeGoogle appears to be developing Gempix 3 Pro, likely Nano Banana 3 Pro, built on Gemini 4 Argon. References point to image-generation and editing upgrades, internal testing, and possible November-December AI announcements.

🚨 AI News | TestingCatalog

@testingcatalog.com

Anthropic tests its next model under Claude Opus 5.5

testingcatalog.comAnthropic tests its next model under Claude Opus 5.5Users report Claude Opus 5.5 is sometimes routing to a newer Anthropic checkpoint, with stronger coding and 3D outputs. Tests suggest fresher training data or a possible Opus 5.6 trial, following recent price cuts.

🚨 AI News | TestingCatalog

@testingcatalog.com

Gemini 4 Argon spotted in A/B testing on Antigravity

testingcatalog.comGemini 4 Argon spotted in A/B testing on AntigravityLeaked A/B tests suggest Google’s unreleased Gemini 4 Argon is quietly appearing behind other model names. Early outputs look stronger than Gemini 3.8 Flash, but the exact checkpoint and real-world performance remain unclear.

🚨 AI News | TestingCatalog

@testingcatalog.com

Anthropic tests Claude Health for web and mysterious "iggy"

testingcatalog.comAnthropic tests Claude Health for web and mysterious "iggy"Anthropic is testing two unreleased Claude interface sections, Arborio and Iggy. Arborio may relate to Claude Health on the web, while Iggy’s purpose is unknown. Neither name nor feature set, timeline, or public launch is confirmed.

🚨 AI News | TestingCatalog

@testingcatalog.com

Anthropic might be developing its own voice models

testingcatalog.comAnthropic might be developing its own voice modelsAnthropic appears to be testing a hidden Claude Voice preview in its internal web environment. The finding follows a new opt-in setting for sharing voice data for training, pointing to possible speech model work or changes to current voice features.

🚨 AI News | TestingCatalog

@testingcatalog.com

OpenAI lists new gpt-rosalind-discovery model in docs

testingcatalog.comOpenAI lists new gpt-rosalind-discovery model in docsOpenAI’s pricing page now lists unannounced gpt-rosalind-discovery beside gpt-rosalind-research at identical rates. The move suggests a broader Rosalind family, likely aimed at biology and drug discovery under trusted access.

🚨 AI News | TestingCatalog

@testingcatalog.com

Gemini 4 Argon hints emerge as Google tests Carbon checkpoint

testingcatalog.comGemini 4 Argon hints emerge as Google tests Carbon checkpointGoogle appears close to releasing Gemini 4 Argon, with Antigravity showing 256K, 512K, and 900K context options. Internal tests also point to Carbon, a separate Gemini 4 checkpoint aimed at advanced coding agents.

🚨 AI News | TestingCatalog

@testingcatalog.com

Microsoft launches Decision-1 model in Foundry

testingcatalog.comMicrosoft launches Decision-1 model in FoundryMicrosoft launched Decision-1, a fast scoring model for fixed-choice tasks. Available in Foundry, it returns calibrated probabilities, posts strong benchmark results, cuts latency and cost, and targets routing, safety, agents and workflow control.

🚨 AI News | TestingCatalog

@testingcatalog.com

How Sabi's brain-reading cap turns thoughts into text

testingcatalog.comHow Sabi's brain-reading cap turns thoughts into textSabi is developing a noninvasive EEG cap that decodes internal speech into text. Using custom non-contact sensors and a neural model trained on large in-house datasets, it aims to make brain-to-computer input practical for daily use.

🚨 AI News | TestingCatalog

@testingcatalog.com

Google prepares new Ultra mode for AI Studio Build

testingcatalog.comGoogle prepares new Ultra mode for AI Studio BuildGoogle is testing an unreleased Ultra mode in AI Studio Build, listed beside Plan, Build, and Security. Early tests suggest deeper reasoning for complex app generation, but its tools, model ties, pricing, and release timing remain unknown.

🚨 AI News | TestingCatalog

@testingcatalog.com

Claude launches Dashboards and Motion in beta

testingcatalog.comClaude launches Dashboards and Motion in betaAnthropic launched Claude Dashboards and Claude Motion in beta, adding live data dashboards and code-based animations. Docs, Slides, and Design are now on all plans, while standalone Claude Design closes December 14.

🚨 AI News | TestingCatalog

@testingcatalog.com

OpenAI rolls out GPT-6 with Intelligent UI in ChatGPT

testingcatalog.comOpenAI rolls out GPT-6 with Intelligent UI in ChatGPTOpenAI is rolling GPT-6 and Intelligent UI across ChatGPT, adding visual, task-shaped answers with charts, forms and tools. Paid tiers get Sol first, Free and Go follow, while Work and Codex models remain unchanged.

🚨 AI News | TestingCatalog

@testingcatalog.com

Microsoft brings hybrid AI and agent controls to Windows

testingcatalog.comMicrosoft brings hybrid AI and agent controls to WindowsMicrosoft outlined a Windows hybrid AI strategy that combines local and cloud models, adds policy-based controls for agents, expands Copilot and developer tooling, and pushes larger on-device models across PCs and enterprise systems.

🚨 AI News | TestingCatalog

@testingcatalog.com

Anthropic launches Claude Haiku 5.5 at lower API prices

testingcatalog.comAnthropic launches Claude Haiku 5.5 at lower API pricesAnthropic launched Claude Haiku 5.5, a faster, lower-cost small model for high-volume tasks. It adds adjustable effort, posts large gains over Haiku 4.5, and is available across major clouds and Anthropic’s platform.

🚨 AI News | TestingCatalog

@testingcatalog.com

Google prepares Antigravity for Gemini 4 Argon and Concierge

testingcatalog.comGoogle prepares Antigravity for Gemini 4 Argon and ConciergeGoogle appears to be preparing Antigravity for Gemini 4 Argon, with labels tied to Ultra plans and upgrade prompts. Internal prototypes point to voice control, auto approvals, and customization, but no release date is confirmed.

🚨 AI News | TestingCatalog

@testingcatalog.com

Mitra adds chat takeover and Mitra Link to its AI assistant

testingcatalog.comMitra adds chat takeover and Mitra Link to its AI assistantMitra is expanding from an AI phone agent into a cross-channel assistant that can join chats, reply from a user’s account, take over workstreams, and connect via Mitra Link to carry conversations across apps and contacts.

🚨 AI News | TestingCatalog

@testingcatalog.com

Mistral launches Large 4 preview with 1 T parameters

testingcatalog.comMistral launches Large 4 preview with 1 T parametersMistral launched public preview of Mistral Large 4, a 1T-parameter multimodal MoE model with 49B active parameters. It targets coding, cyber, finance and law, posts strong benchmark results, and brings API access now with weights due later.

🚨 AI News | TestingCatalog

@testingcatalog.com

Figure plans to release Hark AI assistant this week

testingcatalog.comFigure plans to release Hark AI assistant this weekFigure plans to launch Hark as a proactive personal AI with memory, browser actions, and connected email, calendars, and docs. Early details point to a home screen for tasks, briefs, widgets, and web-based assistant workflows.

🚨 AI News | TestingCatalog

@testingcatalog.com

Aleph Alpha releases open-weight Kolibri with 1M context

testingcatalog.comAleph Alpha releases open-weight Kolibri with 1M contextAleph Alpha released Kolibri, a bilingual open-weight 78.1B MoE model for sovereign government and regulated use. Apache 2.0 licensed, on-prem deployable, it offers 1M-token context, German focus, tool calling, and grounding.

🚨 AI News | TestingCatalog

@testingcatalog.com

Connecting the "dots" between ChatGPT and World Wallet

testingcatalog.comConnecting the "dots" between ChatGPT and World WalletLeaked app screens suggest ChatGPT will add a wallet for agent-led purchases. With OpenAI testing payment flows and World App rebranding to World Money, the timing hints at a possible link, though no public evidence confirms it.

🚨 AI News | TestingCatalog

@testingcatalog.com

Microsoft launches MAI-Transcribe-2-Streaming

testingcatalog.comMicrosoft launches MAI-Transcribe-2-StreamingMicrosoft launched MAI-Transcribe-2-Streaming and two MAI-Voice models, offering real-time transcription, multilingual text-to-speech, low-latency audio generation, and voice cloning for conversational agents.

🚨 AI News | TestingCatalog

@testingcatalog.com

Gemini desktop to get broader Computer Use permissions

testingcatalog.comGemini desktop to get broader Computer Use permissionsGoogle is developing broader Gemini desktop permissions that could let it operate across apps, files, and the web with fewer prompts, while keeping approval for sensitive actions. The feature remains in trusted testing with no launch date.

🚨 AI News | TestingCatalog

@testingcatalog.com

TinyFish opens student and ambassador builder programs

testingcatalog.comTinyFish opens student and ambassador builder programsTinyFish launched free Student and Ambassador programs that pay students and indie developers to build agents on its web APIs. Rewards, credits, referrals, and promotions are tied to approved work and qualified signups.

🚨 AI News | TestingCatalog

@testingcatalog.com

Google unveils Gemini 4 Argon with SOTA score on DeepSWE

testingcatalog.comGoogle unveils Gemini 4 Argon with SOTA score on DeepSWEGoogle introduced Gemini 4 Argon, a reasoning model for long software, enterprise and cyber tasks. It launches first to trusted defenders, then paid API and AI Ultra users, with lower intro pricing and a 1M-token output limit.

🚨 AI News | TestingCatalog

@testingcatalog.com

OpenAI launches Dots agents powered by GPT-6 Astra

testingcatalog.comOpenAI launches Dots agents powered by GPT-6 AstraOpenAI’s Dots are always-on GPT-6 Astra agents that handle ongoing work across apps and devices, retain context, learn user preferences, and operate under permissions, review controls, and enterprise governance.

🚨 AI News | TestingCatalog

@testingcatalog.com

OpenAI launches GPT-6.1 Sol at one-fifth of Astra pricing

testingcatalog.comOpenAI launches GPT-6.1 Sol at one-fifth of Astra pricingOpenAI launched GPT‑6.1 Sol, offering near‑Astra results for coding, computer use and business workflows at about one‑fifth the cost, with cached input at $0.10 per million tokens and stronger accuracy, safety and tool use.

🚨 AI News | TestingCatalog

@testingcatalog.com

Meta launches Muse for Small Business with new connectors

testingcatalog.comMeta launches Muse for Small Business with new connectorsMeta launched Muse for Small Business, an AI agent that connects Meta tools and third-party apps to support planning and operations. It can use business data across services, with approval required before posting, messaging or spending.

🚨 AI News | TestingCatalog

@testingcatalog.com

Anthropic launches Claude Sonnet 5.5 with faster output

testingcatalog.comAnthropic launches Claude Sonnet 5.5 with faster outputAnthropic launched Claude Sonnet 5.5 as a faster, lower-cost upgrade to Sonnet 5 for everyday work. It delivers over 30% faster output, lower task costs, strong benchmark gains, broader platform support, and new cyber safeguards.

🚨 AI News | TestingCatalog

@testingcatalog.com

Manus 2.0 launches with Studio, Cloud Computer and Cue

testingcatalog.comManus 2.0 launches with Studio, Cloud Computer and CueManus 2.0 expands across web, desktop, and mobile with a new agent system, Studio, automations, game and video tools, device control, and Cue personal agents, while reporting lower token use, faster task completion, and reduced cost.

🚨 AI News | TestingCatalog

@testingcatalog.com

ElevenLabs launches Eleven v4 and v4 Turbo voice models

testingcatalog.comElevenLabs launches Eleven v4 and v4 Turbo voice modelsElevenLabs launched Eleven v4 for high-quality scripted speech and v4 Turbo for real-time voice agents, combining multi-speaker control, audio tags, 90+ languages, and low-latency streaming via its products, API, and agents.

🚨 AI News | TestingCatalog

@testingcatalog.com

Anthropic may release Claude Sonnet 5.5 within days

testingcatalog.comAnthropic may release Claude Sonnet 5.5 within daysLeaks suggest Anthropic may launch Claude Sonnet 5.5 as early as next week. Reports point to stronger performance, up to 1M context, and competitive pricing as Anthropic, OpenAI, and Google race to release new flagship AI models.