https://tpsreport.news/news/elevenlabs-music-v2-5-api-launch TPS — Tokens Per Second en 2026-09-13T13:50:56.125Z ElevenLabs Launches Music v2.5, Adds API Access and Free Tier for AI-Generated Songs ElevenLabs, AI Music Generation, ElevenMusic, Music v2.5, API, Suno, Universal Music Group https://tpsreport.news/news/iris-mini-iris-pro-open-weight-search-agents TPS — Tokens Per Second en 2026-09-13T13:06:01.560Z AllSpark's Iris-mini and Iris-pro Top Open-Weight Search Agent Benchmarks search-agents, open-weight-models, Qwen3, AllSpark, benchmarks, reinforcement-learning, agentic-AI https://tpsreport.news/news/gpt-6-astra-vending-bench-drone-bench-results TPS — Tokens Per Second en 2026-09-13T11:05:53.990Z OpenAI's GPT-6 Astra Beats Claude Fable 5.1 Nearly 3-to-1 in Autonomous Business Benchmark, Tops Drone Navigation Tests GPT-6 Astra, OpenAI, Claude Fable, Andon Labs, Vending-Bench, Drone-Bench, AI agents, benchmark https://tpsreport.news/news/amodei-plan-slow-ai-capabilities-pace TPS — Tokens Per Second en 2026-09-12T16:50:38.564Z Anthropic CEO Dario Amodei Proposes Three-Step Plan to Deliberately Slow AI Capability Advances Anthropic, Dario Amodei, AI safety, AI policy, AI governance, existential risk https://tpsreport.news/news/gpt-6-astra-spatial-reasoning-benchmark TPS — Tokens Per Second en 2026-09-12T14:35:47.578Z GPT-6 Astra Beats Ai2's MolmoAct2 on New Robotics Benchmark, Researcher Calls It a 'Step Change' GPT-6 Astra, OpenAI, Ai2, MolmoAct2, robotics, spatial reasoning, benchmark https://tpsreport.news/news/tencent-auk-open-source-speech-model TPS — Tokens Per Second en 2026-09-12T14:05:48.005Z Tencent Open-Sources AuK, a 1.5B-Parameter Speech Generation and Editing Model tencent, text-to-speech, open-source, speech-generation, audio-model, MIT-license https://tpsreport.news/news/ai-reasoning-steps-internal-patterns-study TPS — Tokens Per Second en 2026-09-12T13:50:40.267Z Study Finds AI Models' Reasoning Steps Leave Distinct Fingerprints in Internal Activations interpretability, chain-of-thought, AI safety, reasoning models, KAIST, Naver AI Lab https://tpsreport.news/news/anthropic-threat-report-claude-misuse-war-surveillance TPS — Tokens Per Second en 2026-09-12T12:50:42.135Z Anthropic Report: Claude Was Used to Target US Navy Ships, Build Missiles, and Track Uyghurs Anthropic, Claude, AI safety, national security, surveillance, threat intelligence, misuse https://tpsreport.news/news/google-timesfm-3-multivariate-forecasting-model TPS — Tokens Per Second en 2026-09-12T09:35:58.330Z Google Releases TimesFM-3, a 330M-Parameter Model That Forecasts Sales Using Weather and Discount Data Google, TimesFM-3, time series forecasting, forecasting AI, Google Research, model release https://tpsreport.news/news/deepseek-v4-1-flash-encoder-decoder-architecture TPS — Tokens Per Second en 2026-09-12T06:05:54.641Z DeepSeek Ships V4.1-Flash With Novel Encoder-Decoder Architecture, Cuts KV Cache to 1/8 of Predecessor DeepSeek, open-weight models, model architecture, encoder-decoder, inference efficiency, vision models, MoE https://tpsreport.news/news/schematron-v2-turbo-html-to-json-extraction TPS — Tokens Per Second en 2026-09-12T01:51:06.705Z Inference.net Launches Schematron V2 Turbo, a 3B-Parameter Model for High-Volume HTML-to-JSON Extraction Inference.net, Schematron, HTML to JSON, structured extraction, small language models, OpenRouter https://tpsreport.news/news/schematron-v2-small-html-json-extraction TPS — Tokens Per Second en 2026-09-12T01:50:42.959Z Inference.net Releases Schematron V2 Small, a 3B-Parameter Model for HTML-to-JSON Extraction Inference.net, Schematron, HTML extraction, JSON extraction, small language models, OpenRouter https://tpsreport.news/news/gpt-terra-sol-astra-openrouter-unconfirmed-1 TPS — Tokens Per Second en 2026-09-12T00:22:01.616Z Three Unverified 'GPT' Codenames — Terra, Sol, Astra — Surface on OpenRouter Under OpenAI's Name OpenAI, OpenRouter, stealth model, rumor, GPT, unconfirmed release, roundup, trending https://tpsreport.news/news/perplexity-gpt-6-astra-engineering-systems TPS — Tokens Per Second en 2026-09-12T00:21:34.551Z Perplexity Says It Runs End-to-End Engineering Systems on OpenAI's GPT-6 Astra OpenAI, Perplexity, GPT-6, Astra, autonomous coding, case study https://tpsreport.news/news/perplexity-openai-astra-engineering-tasks TPS — Tokens Per Second en 2026-09-12T00:20:42.907Z Perplexity Deploys OpenAI's Astra Model for Autonomous Code and Systems Management OpenAI, Perplexity, Astra, AI coding agents, production monitoring, case study https://tpsreport.news/news/gpt-terra-sol-astra-openrouter-codenames TPS — Tokens Per Second en 2026-09-12T00:06:29.614Z Three Unconfirmed 'GPT' Codenames — Terra, Sol, Astra — Appear on OpenRouter Under Unofficial OpenAI Namespace OpenAI, OpenRouter, unconfirmed, rumor, GPT, roundup, trending https://tpsreport.news/news/augment-software-factory-4-5x-developer-output TPS — Tokens Per Second en 2026-09-12T00:06:05.702Z Augment Code Claims 4.5x Developer Output Increase From Internal 'Software Factory' of AI Agents Augment Code, AI agents, software development, coding agents, Cosmos, SDLC automation, developer productivity