https://tpsreport.news/news/open-model-roundup-inkling-laguna-kimi-k3 TPS — Tokens Per Second en 2026-08-02T13:21:58.950Z Open Model Race Intensifies: Thinking Machines, Poolside, Moonshot Ship Competing Frontier Releases open models, Thinking Machines, Poolside, Moonshot AI, Kimi K3, Tencent, DeepSeek, MoE https://tpsreport.news/news/open-artifacts-23-laguna-inkling-kimi-k3 TPS — Tokens Per Second en 2026-08-02T13:21:33.172Z Open Model Race Intensifies: Thinking Machines, Tencent, Poolside, Moonshot Ship Frontier-Class Releases in Same Week open models, open weights, Thinking Machines, Tencent, Poolside, Moonshot AI, Kimi K3, DeepSeek https://tpsreport.news/news/openai-presence-enterprise-ai-agents TPS — Tokens Per Second en 2026-08-02T13:21:01.373Z OpenAI Launches Presence, an Enterprise Service to Push AI Agents Into Production OpenAI, AI agents, enterprise AI, Presence, production deployment, EU AI Act https://tpsreport.news/news/meta-ai-memory-agent-long-tasks TPS — Tokens Per Second en 2026-08-02T13:06:04.773Z Meta AI Pairs a Second 'Memory Agent' With Coding Agents, Lifts Terminal-Bench Score From 38% to 46% Meta AI, AI agents, memory, agentic AI, Claude, Terminal-Bench, research https://tpsreport.news/news/claude-opus-5-prompt-to-3d-game-generation TPS — Tokens Per Second en 2026-08-02T09:06:15.572Z Anthropic's Claude Opus 5 Generates Full 3D Games From a Single Text Prompt, No Assets Required Claude Opus 5, Anthropic, 3D generation, game development, Three.js, procedural generation, AI code generation https://tpsreport.news/news/openai-astra-ten-math-proofs TPS — Tokens Per Second en 2026-08-01T20:51:07.071Z OpenAI Claims Internal Astra Model Solved 10 Decade-Old Math Problems for Under $2,000 Each mathematics, openai, astra, reasoning-models, lean4, theoretical-computer-science https://tpsreport.news/news/deepseek-v4-flash-latest-openrouter TPS — Tokens Per Second en 2026-08-01T18:20:36.839Z DeepSeek Launches 'V4 Flash Latest' Alias with 1M+ Token Context on OpenRouter DeepSeek, DeepSeek V4, OpenRouter, long context, pricing, changelog https://tpsreport.news/news/ai-cracks-unsolved-math-problems-mathematicians-react TPS — Tokens Per Second en 2026-08-01T16:20:40.458Z OpenAI Model Disproves 78-Year-Old Erdos Conjecture, Triggering Mixed Reaction From Mathematicians mathematics, OpenAI, Erdos problems, Terence Tao, AI research, benchmarks, FrontierMath https://tpsreport.news/news/ai-coding-agents-research-software-validation-report TPS — Tokens Per Second en 2026-08-01T14:36:04.442Z OpenAI Field Report: Coding Agents Speed Up Research Software 60x But Can't Verify Scientific Correctness AI coding agents, research software, OpenAI, Codex, Claude Code, scientific computing, software validation, genomics https://tpsreport.news/news/bytedance-seedance-2-5-30-second-video-audio TPS — Tokens Per Second en 2026-08-01T13:50:38.907Z ByteDance's Seedance 2.5 Generates 30-Second AI Video Clips With Synced Audio ByteDance, Seedance, AI video generation, text-to-video, multimodal AI https://tpsreport.news/news/openai-astra-model-family-long-running-tasks TPS — Tokens Per Second en 2026-08-01T08:35:37.895Z OpenAI Reportedly Developing 'Astra' Model Family for Multi-Day Autonomous Problem-Solving OpenAI, Astra, agentic AI, long-horizon reasoning, AI regulation, GPT-6, multi-agent systems https://tpsreport.news/news/deepseek-v4-flash-0731-post-training-update TPS — Tokens Per Second en 2026-08-01T01:51:01.454Z DeepSeek V4-Flash 0731 Update Jumps Terminal-Bench Score by 25.8 Points With No Architecture Change DeepSeek, DeepSeek V4-Flash, open weights, post-training, pricing, AI security, benchmarks, Anthropic https://tpsreport.news/news/gemini-app-ai-studio-mobile-cancelled TPS — Tokens Per Second en 2026-07-31T23:50:43.200Z Google Cancels Standalone AI Studio Mobile App, Shifts App-Building Into Gemini App Instead Gemini, Google DeepMind, AI Studio, mobile apps, generative UI, product cancellation https://tpsreport.news/news/gemini-2-5-pro-gemini-3-flash-deprecated-github-copilot TPS — Tokens Per Second en 2026-07-31T20:21:12.521Z GitHub Copilot Deprecates Gemini 2.5 Pro and Gemini 3 Flash Across All Experiences GitHub Copilot, Gemini 2.5 Pro, Gemini 3 Flash, Google DeepMind, deprecation, changelog https://tpsreport.news/news/gemini-robotics-2-google-deepmind-vla-model TPS — Tokens Per Second en 2026-07-31T18:35:47.038Z Google DeepMind Launches Gemini Robotics 2, a Single VLA Model for Arms to Humanoids Google DeepMind, Gemini Robotics, robotics, vision-language-action, humanoid robots, embodied AI https://tpsreport.news/news/thinking-machines-inkling-small-efficient-reasoning-model TPS — Tokens Per Second en 2026-07-31T17:50:54.175Z Thinking Machines Releases Inkling Small, a 12B-Active-Parameter Model That Beats Its Larger Predecessor on Key Benchmar Thinking Machines, Inkling Small, open-weights, reasoning model, Mira Murati, mixture-of-experts, Apache 2.0