Skyler Miao said MiniMax M2.7 open weights are due in roughly two weeks, with updates tuned for agent tasks. Separate replies also confirm multimodal M3, so local-stack builders should watch both the drop and the benchmark setup.
The clearest hard signal is the M2.7 weights timeline. In the screenshot embedded in open-weights post, Skyler Miao says "M2.7 open weights coming in ~2 weeks" and adds that the team is "still actively iterating." That is stronger than rumor, but it is still a social-post commitment rather than a formal launch note with license, parameter count, or deployment requirements.
MiniMax separately confirmed a multimodal follow-on. The M3 vision reply shows a user asking whether MiniMax models will have vision, and Miao answers: "Sure, in M3." MiniMax's own account later amplified that same exchange in MiniMax repost, which makes the M3 multimodal plan look intentional rather than a stray reply. One context post also claims M3 will be a "big 1T model" speculation thread, but that size figure is not confirmed by MiniMax in the provided evidence.
The interesting engineering detail is that MiniMax is positioning M2.7 as more than a generic open release. In benchmark repost, the company says it performed "extensive optimizations" and "established a dedicated benchmark" for this workload. The attached MM-ClawBench chart shows M2.7 at 62.7 versus M2.5 at 57.6, slightly above Gemini 3.1 Pro at 61.8, below Claude Sonnet 4.6 at 64.2, and behind Opus 4.6 and GPT-5.4.
The OpenClaw mention matters because it points to agentic, tool-using evaluation rather than pure chat vibes. The screenshot in open-weights post says the newest revision is "noticeably better on OpenClaw," suggesting the team is still optimizing around autonomous coding or browser-style task completion before release. Community reaction is already translating that into deployment terms: home-run reaction calls M2.7 "the best model you can realistically run at home" and says the author already runs M2.5 locally. Meanwhile, HF org page shows MiniMax has a public Hugging Face presence with prior model and research activity, a practical signpost for where engineers may want to watch for the actual drop.
OpenHands introduced EvoClaw, a benchmark that reconstructs milestone DAGs from repo history to test continuous software evolution instead of isolated tasks. The first results show agents can clear single tasks yet still collapse under regressions and technical debt over longer runs.
releaseOpenClaw shipped version 2026.3.22 with ClawHub, OpenShell plus SSH sandboxes, side-question flows, and more search and model options, then followed with a 2026.3.23 patch. Teams get a broader plugin surface, but should patch quickly and review plugin trust boundaries as the ecosystem grows.
releaseCursor shipped Instant Grep, a local regex index built from n-grams, inverted indexes, and Bloom filters that drops large-repo searches from seconds to milliseconds. Faster candidate retrieval shortens the coding-agent loop, especially when ripgrep-style scans become the bottleneck.
breakingChatGPT now saves uploaded and generated files into an account-level Library that can be reused across conversations from the web sidebar or recent-files picker. It removes repetitive re-uploading and makes past PDFs, spreadsheets, and images part of a persistent working context.
breakingEpoch AI says GPT-5.4 Pro elicited a publishable solution to one 2019 conjecture in its FrontierMath Open Problems set, with a formal writeup planned. Treat it as an early milestone worth reproducing, not blanket evidence that frontier models can already automate math research.
time to follow: huggingface.co/MiniMaxAI
M2.7 open weights coming in ~2 weeks. still actively iterating just updated a new version on yesterday — noticeably better on OpenClaw.
MiniMax M2.7 open weights is coming in ~2 weeks
M2.7 open weights coming in ~2 weeks. still actively iterating just updated a new version on yesterday — noticeably better on OpenClaw.
MiniMax M3 confirmed, multimodal too?
Sure, in M3.
Yes, we performed extensive optimizations and even established a dedicated benchmark for it.
night and day difference between minimax 2.5 and 2.7 on openclaw. i'm very bullish on minimax now. i use it over other models!!!
