Qwen3.8 Max tops agentic index, OpenAI opens up
Two frontier moves in one day: Alibaba's Qwen3.8 Max is the new #1 on Artificial Analysis's agentic index, while OpenAI shipped a sharper GPT-5.6 Sol and unlimited text chats for free users. Plus: the xAI pollution fight, a YC-bound agent runtime, and a record defense-tech round.
Qwen3.8 Max, Alibaba's latest flagship, now ranks as the best overall model on Artificial Analysis's agentic index — the first open-weight model to top it. The company calls Qwen3.8-Max "the first open-weight model at Max scale" and its benchmark table shows it leading on tests like PaperBench (93.0) while holding its own against Opus 4.8, Fable 5, and GPT-5.6 Sol on coding and agentic work. Open-weight models keep closing the gap with closed frontier labs — the interesting question is whether pricing and ecosystem adoption follow.
OpenAI is upgrading GPT-5.6 Sol for Plus/Pro users and giving Free and Go users unlimited text chats on GPT-5.6 Luna. The refreshed Sol is tuned for more focused, fact-reliable answers with a new effort slider; Luna becomes the free default this week, with a "Think" button for harder questions arriving next week. OpenAI says internal evals show factual errors dropped ~62% for Luna and ~68% for Sol versus GPT-5.5 Instant. Removing text limits for a base that recently crossed 1 billion weekly users is the clearest sign yet that OpenAI competes on reach, not just raw capability.
xAI is under a Clean Air Act lawsuit over an unpermitted gas-turbine plant powering its Colossus 2 data center in Southaven, Mississippi. SELC and Earthjustice, on behalf of the NAACP, sued over 27 unpermitted methane turbines that can emit over 1,700 tons of NOx per year, and the case has drawn Senate scrutiny. Data-center power is becoming the industry's biggest externality — expect more of these fights as AI buildout outruns permitting regimes.
Herdr, the open-source runtime for coding agents, is joining Y Combinator — and staying open. The team says the terminal-native runtime remains open source through the accelerator. It's another sign the agent-infrastructure layer is consolidating into funded, standardized tooling rather than one-off scripts.
Hadrian, which builds AI-powered factories for defense manufacturing, raised $1.37 billion at a nearly $8 billion valuation. The Series D, reported by Axios and TechCrunch, roughly quadruples the company's January valuation amid surging demand for defense production. The bet: the bottleneck in defense is manufacturing throughput, and software-controlled factories are the fix.
What to watch: OpenAI's unlimited free text chats roll out next week — and whether Qwen's open-weight lead forces a pricing response from closed labs.
If an open-weight model tops the agentic benchmarks, is the closed-lab advantage finally over? Tell us in the comments.
