MiMo-V2.6-Pro, Claude Opus 5.5, GPT-6 Luna, and GPT-6 Sol account for 11 new points on Artificial Analysis’s Intelligence Index versus cost-per-task Pareto frontier, across different reasoning efforts. Claude Opus 5.5 (max with fallback) scores 58 at $5.98 per task, the highest result on the index.
At Climate Week NYC, Ai2 and GlobalFishWatch announced a partnership on how AI and AI agents are used in ocean monitoring and enforcement. Ai2 describes the effort as purpose-built, open, and accountable to the people who use it.
The discount applies to both Alibaba_Qwen models on Together AI serverless and runs through September 30. Together AI recommends Qwen3.7-Max for long-horizon agent work and Qwen3.8-Flash for high-volume, cost-sensitive workloads.
Polymarket says Alabama has begun teaching AI literacy in schools, with instruction starting in kindergarten.
Anthropic says biologists use Claude to propose hypotheses and candidate biological systems, while its scientists review those ideas and perform all laboratory work. The company is inviting research-question proposals in genomics and other fields.
Anthropic says the enzyme’s gene sits beside a long array of repeating DNA that looks somewhat similar to CRISPR. It says the system’s function is not yet understood, while the few known systems sharing its features can all cut, copy, and paste DNA.
Victor M says the model turns long documents into small page images to save tokens, then retrieves the full text of only the pages relevant to a question.
Model Vault is now available in Canada, with auto-scaled workloads, single tenancy, and completely private Cohere models. Cohere says the setup can lower total cost of ownership.
News from Google says the open-weight, Gemma-based models for medical text and images have been downloaded more than 10 million times since release. It describes them as a foundation for developers, researchers, and public health organizations building tools for triage, diagnostic screenings, and other uses.
Polymarket says Anthropic’s new wet lab made its first discovery after nearly 1,000 Claude agents worked autonomously for 21 hours.
Polymarket reported that the Senate was scheduled to hold a hearing at 2:30 p.m. ET on September 23, 2026, examining Flock Safety’s nationwide AI surveillance network and how Americans’ license plate data is accessed and shared.
GPT-Live-1 carries the conversation as the Raspberry Pi sends research and tool tasks to GPT-5.6 Luna through the Responses API. A local renderer converts scenes to 128 × 64 pixels, and Sid can keep talking while that work continues in the background.
Weights, code, a fine-tuning recipe, benchmarks, and reproducible examples are being released. Black Forest Labs says the world action model ranks first on RoboLab, 6.1 percentage points ahead of the previous best open model, with 56% fewer parameters and up to 3.95 times faster running speed.
Command Code has added 12 models, including Mimo V2.6 Pro Ultraspeed, GLM 5.3 FlashX, Qwen 3.8 Omni Flash, and Step 5 Preview. DeepSeek V4.1 Flash is included with extended usage, and Jev is free.
OpenRouter says the issue that took Space Bunny Alpha offline is resolved and the model is serving again.
On Artificial Analysis’s Pronunciation Robustness benchmark, Gemini 3.8 Flash TTS scores 89.5%, ahead of Gemini 3.1 Flash TTS, SpaceXAI TTS, and Gemini 3.8 Flash-Lite TTS. Leadership on individual pronunciation categories is split among several models.
Polymarket says the system is designed to run up to 380,000 sandboxes simultaneously and contain "agent misbehavior."
OpenAI says ChatGPT Voice can use plugins such as email, calendar, and Slack, run on GPT-6 Astra, Sol, and Luna, and work in ChatGPT Work on the web and mobile. The update is rolling out globally today in the latest version of the app.
Arena.ai reports that OpenAI’s GPT-6 Sol (Max) scored 1,689 points and placed fourth in Code Arena: WebDev, at a blended price of $8 per million tokens. It is 72 points above GPT-5.6 Sol (xHigh), while Arena.ai says GPT-6 Luna’s score is still pending.
OpenRouter lists Recraft V4.1 Flash at $0.007 per image and 1.3 seconds per image. It says the output still looks like Recraft for photography, portraits, complex scenes, logos, and early concepts.
