https://tpsreport.news/news/reka-rho-1-19b-omni-model-video-robot-control TPS — Tokens Per Second en 2026-10-05T18:35:53.555Z Reka AI releases Rho-1, a 19B-parameter omni-model for text, image, video and robot control Reka AI, Rho-1, omni-model, multimodal, world models, robotics, video generation, research preview https://tpsreport.news/news/openai-textgrain-text-watermark-eu-chatgpt-api-opt-in TPS — Tokens Per Second en 2026-10-05T17:50:45.730Z OpenAI to watermark ChatGPT and Codex text in the EU with textGrain; API watermarking is opt-in worldwide OpenAI, ChatGPT, textGrain, watermarking, EU AI Act, Codex, Anthropic, SynthID https://tpsreport.news/news/claude-code-bedrock-govcloud-opus-5-5-sonnet-5-5 TPS — Tokens Per Second en 2026-10-05T17:36:08.386Z Claude Opus 5.5 and Sonnet 5.5 now on Amazon Bedrock in AWS GovCloud (US), with Claude Code support Claude Code, Amazon Bedrock, AWS GovCloud, Anthropic, FedRAMP, ITAR, Claude Opus 5.5, Claude Sonnet 5.5 https://tpsreport.news/news/aws-sagemaker-aws-ai-ml-agent-skill-inference-optimization TPS — Tokens Per Second en 2026-10-05T17:35:46.338Z AWS ships aws-ai-ml skill so Kiro, Claude Code and Codex can benchmark SageMaker inference endpoints Amazon SageMaker, AWS, agent skills, MCP, inference optimization, Kiro, Claude Code, Codex https://tpsreport.news/news/github-reviewbench-open-benchmark-ai-code-review-1 TPS — Tokens Per Second en 2026-10-05T16:06:31.822Z GitHub launches ReviewBench, an open benchmark for AI code review agents built on real pull requests ReviewBench, GitHub, GitHub Copilot, code review, benchmark, AI agents, evaluation https://tpsreport.news/news/github-reviewbench-open-benchmark-ai-code-review TPS — Tokens Per Second en 2026-10-05T16:06:09.812Z GitHub launches ReviewBench, an open benchmark for AI code review agents built on real pull requests ReviewBench, GitHub, GitHub Copilot, code review, benchmark, AI agents, pull requests, evaluation https://tpsreport.news/news/amazon-quick-resource-migrator-mcp-server-cross-account TPS — Tokens Per Second en 2026-10-05T16:05:51.791Z AWS releases open-source MCP server to automate cross-account promotion of Amazon Quick agents Amazon Quick, Amazon Bedrock AgentCore, MCP, AWS, Model Context Protocol, enterprise AI, agent governance, DevOps https://tpsreport.news/news/reflection-open-weight-model-nvidia-axios-report TPS — Tokens Per Second en 2026-10-04T13:05:47.399Z Nvidia-backed Reflection to release open-weight model soon, Axios reports Reflection, open-weight, Nvidia, Misha Laskin, AI factory, Shinsegae, Nebius, open-source AI https://tpsreport.news/news/google-rrsi-self-improving-agent-harness-overfitting TPS — Tokens Per Second en 2026-10-04T12:50:45.140Z Google's RRSI cuts overfitting in self-improving agents: up to 4.7-point gains on unseen tasks with ~30% fewer tokens Google, agents, agent harness, self-improvement, RRSI, overfitting, research, Claude Opus 4.8 https://tpsreport.news/news/aleph-alpha-benchmark-chinese-ai-models-sensitive-topics TPS — Tokens Per Second en 2026-10-04T09:05:47.407Z Aleph Alpha benchmark: Chinese AI models balanced on just 17-41% of 967 sensitive-topic prompts Aleph Alpha, benchmark, censorship, Qwen 3.6, DeepSeek V4 Pro, Kimi, Nemotron Cascade 2, Claude Sonnet 5 https://tpsreport.news/news/gemini-free-tier-flash-lite-only-ai-plus-loses-pro TPS — Tokens Per Second en 2026-10-04T07:35:45.109Z Google limits free Gemini users to Flash-Lite and removes Pro from the $4.99 AI Plus plan Google, Gemini, Gemini Flash-Lite, Gemini Pro, Google AI Plus, Google AI Pro, Google AI Ultra, pricing https://tpsreport.news/news/microsoft-thinkingbox-agent-benchmark-backend-state TPS — Tokens Per Second en 2026-10-03T23:05:50.243Z Microsoft's ThinkingBox: Claude Opus 5.5 passes all 20 runs on just 241 of 507 stateful agent tasks Microsoft, Hugging Face, ThinkingBox, agent benchmark, OpenEnv, MCP, Claude Opus 5.5, Kimi-K3 https://tpsreport.news/news/gemini-app-model-access-limits-free-ai-plus-deep-think-ai-pro TPS — Tokens Per Second en 2026-10-03T21:05:45.709Z Google limits free Gemini app users to Flash-Lite from Oct. 9; AI Pro adds Deep Think Gemini, Google, Deep Think, Gemini 3.1 Pro, Gemini Flash, Flash-Lite, AI Plus, AI Pro