swyx (@swyx)

DeepSeek v4가 벤치마크 점수 경쟁보다 긴 컨텍스트 효율성 기술(CSA, HCA, mHC 등)과 낮은 비용으로 SOTA 수준의 성능을 보여주며 주목받고 있다. 추론 최적화 비용을 과시하지 않고도 경쟁력을 입증한 점이 인상적이다.

https://x.com/swyx/status/2049569820728078533

#deepseek #llm #longcontext #efficiency #aimodel

swyx 🇸🇬 (@swyx) on X

IMO DeepSeek v4 demonstrated utter confidence and competence by not benchmaxxing, not focusing on some BS final run cost, not even spending inference-optimal compute. just showed up, demonstrated SOTA long context efficiency techniques (CSA, HCA, mHC, flash at 8% cost of pro,

X (formerly Twitter)

DeepSeek-V4-Pro API 价格跳水,官方限时2.5折优惠

DeepSeek V4-Pro API prices are now 25% off until May 5, 2026. Input prices start at $0.25 per million tokens.

#DeepSeekV4Pro, #APIPriceDrop, #AIModel, #TechDeals, #DeepSeek

https://newsletter.tf/deepseek-v4-pro-api-price-discount-may-5-2026/

DeepSeek V4-Pro API prices have dropped significantly, with a 25% discount available until May 5, 2026. Input prices are now as low as $0.25 per million tokens.

#DeepSeekV4Pro, #APIPriceDrop, #AIModel, #TechDeals, #DeepSeek
https://newsletter.tf/deepseek-v4-pro-api-price-discount-may-5-2026/

DeepSeek V4-Pro API price drops to 25% off until May 5, 2026

DeepSeek V4-Pro API prices are now 25% off until May 5, 2026. Input prices start at $0.25 per million tokens.

NewsletterTF

Firefox Exposed: AI Model Uncovers 271 Zero-Day Vulnerabilities

Meet the AI model that just supercharged Firefox security, uncovering a whopping 271 zero-day vulnerabilities that have now been squashed in the latest update to Firefox 150. This game-changing collaboration between Firefox and Anthropic's cutting-edge tools has made the browser safer than ever.

https://osintsights.com/firefox-exposed-ai-model-uncovers-271-zero-day-vulnerabilities?utm_source=mastodon&utm_medium=social

#ZeroDay #Firefox #AiModel #VulnerabilityManagement #BrowserSecurity

Firefox Exposed: AI Model Uncovers 271 Zero-Day Vulnerabilities

Discover how Firefox fixed 271 zero-day vulnerabilities with AI, learn more about their security efforts and upgrade to the latest version now for enhanced protection.

OSINTSights
Distracting your domme during ranked? Bold. #BDSM #AIModel
#Nvidia launched #Nemotron 3 Nano Omni, a 30 billion parameter #AImodel that unifies text, vision, and speech. The model, using a mixture-of-experts architecture, offers low latency and high flexibility, enabling faster and smarter #agenticAI applications. It is available on Hugging Face, OpenRouter, and build.nvidia.com as an Nvidia NIM microservice. https://siliconangle.com/2026/04/28/nvidia-introduces-nemotron-3-nano-omni-vision-speech-powerful-agentic-ai-use/?eicker.news #tech #media #news

Vals AI (@ValsAI)

Kimi K2.6가 Vals Index에서 새로운 1위 오픈웨이트 모델로 발표됐다. 전체 순위는 8위이며, 많은 폐쇄형 모델들과 비슷한 성능을 더 낮은 비용으로 제공해 주목된다.

https://x.com/ValsAI/status/2049256017402433991

#kimi #openweight #llm #benchmark #aimodel

Vals AI (@ValsAI) on X

Results on the remainder of our benchmarks are in for Kimi K2.6, the new #1 open-weight model on the Vals Index (#8 overall) Kimi K2.6 is competitive with many closed-weight models, at a fraction of their price.

X (formerly Twitter)

AI Leaks and News (@AILeaksAndNews)

OpenAI의 차기 모델 GPT-5.6이 내부 codex 로그에서 발견되었다. 어떤 모델이 사용됐는지 검증하는 기록을 통해 확인된 것으로, ChatGPT 제작사의 다음 세대 AI 모델 출시 가능성을 시사한다.

https://x.com/AILeaksAndNews/status/2049191597242187912

#openai #gpt5.6 #llm #codex #aimodel

AI Leaks and News (@AILeaksAndNews) on X

OpenAI’s GPT-5.6 has been spotted The upcoming AI model from the ChatGPT maker was found in internal codex logs that verify which model was being used Are you ready for another big jump in AI capabilities from OpenAI?

X (formerly Twitter)

0xMarioNawfal (@RoundtableSpace)

마이크로소프트가 60분 분량의 오디오를 한 번에 전사할 수 있는 오픈소스 AI 모델을 공개했다. 이 모델은 화자 추적, 긴 컨텍스트 처리, 50개 이상 언어 지원을 제공해 기존 음성 AI 도구의 청킹 문제를 줄이는 데 도움이 된다.

https://x.com/RoundtableSpace/status/2048881120897302681

#microsoft #opensource #speechrecognition #audiotranscription #aimodel

0xMarioNawfal (@RoundtableSpace) on X

MICROSOFT JUST DROPPED AN OPEN SOURCE AI MODEL THAT CAN TRANSCRIBE 60 MINUTES OF AUDIO IN ONE SHOT. The model handles speaker tracking, long context, and 50+ languages without the chunking issues that break most voice AI tools.

X (formerly Twitter)

Ivan Fioravanti ᯅ (@ivanfioravanti)

MiniMax M3가 곧 출시된다는 예고성 트윗입니다. 구체적 기능이나 성능은 공개되지 않았지만, 신규 AI 모델 공개가 임박했음을 알리는 내용이라 개발자 관심도가 높습니다.

https://x.com/ivanfioravanti/status/2048811533367402559

#minimax #m3 #llm #aimodel #release

Ivan Fioravanti ᯅ (@ivanfioravanti) on X

MiniMax M3 is near!

X (formerly Twitter)