Model launchesAlibaba launches Qwen3.7-Max: faster, stronger, but closed weightsAlibaba introduced Qwen3.7-Max, a faster and more capable text-only large language model designed for long-running agentic work and coding.4 min read
Model launchesCorrection: Claude Opus 4's approximately threefold speed increase is dated to May 2025According to the statement, Claude Opus 4's approximately 3× average speed increase is dated to May 2025, not May 2024.1 min read
Model launchesMicrosoft unveils MAI-Thinking-1; benchmark comparisons spark debateAt Build, Microsoft introduced MAI-Thinking-1, a reasoning model with 35 billion active parameters, about 1 trillion total parameters and a 256K token context window.2 min read
Model launchesMeta plans public release of Muse Spark API later this monthMeta has been testing the Muse Spark API with a few initial partners and plans to make it publicly available later this month, a company spokesperson told Reuters.2 min read
Model launchesOpenAI introduces expanded capabilities of the GPT-Rosalind model series in the life sciencesOpenAI announced new capabilities for the GPT-Rosalind model series, which combines GPT-5.5’s agentic coding and tool-use features with scalable intelligence for life-sciences research.1 min read
Model launchesELTE’s Racka-4B: a parameter-efficient Hungarian reasoning language modelEötvös Loránd Tudományegyetem (ELTE) kutatói a Komondor HPC infrastruktúrán fejlesztették a Racka-4B modellt, amely magyar nyelvű feladatokban 8 milliárd paraméteres modellekkel mérhető teljesítményt nyújt, miközben gyorsabb azoknál.3 min read
Model launchesChan Zuckerberg Biohub open-sources ESM Atlas and ESMFold2 protein resourceChan Zuckerberg Biohub has published ESM Atlas, a database of 1.1 billion predicted protein structures, together with the underlying ESMFold2 model under a fully open-source, no-commercial-restrictions licence.2 min read
Model launchesELTE researchers release Racka‑4B: a fast Hungarian-tuned reasoning language modelResearchers at Eötvös Loránd University’s Department of Artificial Intelligence, in cooperation with the National Laboratory for Digital Heritage, have developed Racka‑4B, the first large Hungarian-tuned reasoning language model.3 min read
Model launchesAnthropic releases Claude Opus 4.8 with bugfix-focused improvements, not a major leapAnthropic launched Claude Opus 4.8 forty-two days after Opus 4.7, the shortest interval between releases in the company’s history.2 min read
Model launchesAnthropic unveiled the Opus 4.8 model — a return to the leading position?Anthropic recently released the Opus 4.8 large language model, which has received positive feedback regarding performance and reliability.1 min read
Model launchesGoogle releases faster Gemini 3.5 Flash multimodal model at higher token priceGoogle introduced Gemini 3.5 Flash at Google I/O 2026, a mid-tier multimodal model that prioritizes speed and agentic capabilities but comes at roughly three times the per-token cost of its predecessor.5 min read
Model launchesClaude Opus 4.8 available on the Cursor platform, with better performanceClaude Opus 4.8 has been made publicly available on the Cursor interface; according to CursorBench it operates more efficiently than Opus 4.7 and has proven more persistent on harder tasks.1 min read
