Arena.ai says Claude Opus 5.5 (Max) leads its Code Arena WebDev ranking at a blended $16 per million tokens, 60% below the price of the next-ranked model, GPT-6 Astra (Max). Scores are based on live use by Arena.ai’s global community.
WIRED reports that Australia is investigating whether OpenAI broke the law after an AI agent hacked into the country’s health statistics portal, calling it the first widely known case of an AI agent hacking a government website.
A paper from Google and colleagues describes Harness-Zero, which uses an optimized agent harness only during training. After that harness is removed, macro task success rises from 23.3% to 44.3%, above the 41.7% reached by the base model with the harness attached.
GMI Cloud is offering the Hy Image 3.5 preview at $0.024 per image, a rate it says is 8.8x cheaper than GPT Image 2 and 5.6x cheaper than Nano Banana Pro.
Tencent Hy Translation, powered by Hy-MT2, covers 33 languages, including five Chinese minority languages and dialects, and supports voice and photo translation fully offline on device. Tencent Hy says it is already live in 12 countries and regions.
AI at Meta says raters preferred Muse Realtime Avatar overall after 2–3 minute live calls against two commercial avatar systems, comparing visual quality, sync, character consistency, and mannerisms.
AI at Meta says finkd unveiled Muse Realtime Avatar, real-time embodiment technology that turns Muse Realtime Voice into expressive, interactive avatars, starting with real-time conversations in Muse.
M. Adel Alhashemi released opencode-gpt-live 0.1.1, a plugin for real-time voice calls with an OpenCode session. It uses an existing ChatGPT sign-in and is offered under the MIT license, with GPT-Live handling speech while the session does the work.
OpenEvidence says it announced a partnership with AnthropicAI during the UN General Assembly to make OpenEvidence available free of charge to physicians in low- and middle-income countries.
Supporting plugins in OpenClaw 2026.9.6 can use an optional model for structured choices, separate from chat. TypeSafe Jev sends supplied information to its hosted API and incurs normal charges, while ONNX offers local CPU choices. Decision models are off by default.
The 2026.9.6 release also lists managed updates, restart recovery, 30-day usage, a GitHub reader, remote files, memory and skills, live meeting notes, and Jev with decision models. OpenClaw cites 2,614 pull requests and 351 contributors.
Arena.ai says Claude Opus 5.5 (Max) delivers top performance at a blended $16 per million tokens, and describes that price-performance point as reshaping the Pareto frontier.
Arena.ai places Claude Opus 5.5 (Max) first in Code Arena WebDev at 1,818 points, 26 ahead of GPT-6 Astra (Max) and 126 above Opus 5 (Max). Claude says the model matches Claude Fable 5.1 on most tasks and costs 40% less to run than Opus 5.
Tencent Hy says learning-rate retuning can preserve learning per response across GRPO and PPO over a bounded range of batch sizes. On fixed hardware, larger batches improve PPO generation-stage throughput by up to 2.29×, and its best measured GRPO configuration reaches the same validation target in 29% less time.
Command Code is offering Space Bunny Alpha, which it calls a stealth model, at no charge on all plans. The model has 1M context and is multimodal.
Waabi said the system had not previously practiced the 300-mile drive in simulation and did not rely on new real-world data, calling the run a first.
ClaudeDevs says officially available Claude Code cloud sessions are billed on Pro or Max plans like the rest of Claude Code. Existing subscribers get an optional one-time credit of $100 on Pro or $250 on Max, which those sessions spend before normal plan usage.
In Suno Studio, arming a track and pressing record captures a performance straight onto the timeline.
Artificial Analysis says Claude Opus 5.5 uses 15.6 million tokens per Coding Agent Index task, compared with 11.4 million for Opus 5. Cached input rises from 10.9 million to 14.6 million, and output more than doubles from 137,000 to 333,000.
At max effort in Claude Code, Artificial Analysis measures Opus 5.5 at 66, above Opus 5 at 60 and Claude Fable 5.1 at 62. Its cost per task is $13.04, higher than Opus 5’s $10.79 despite lower token prices.
