Models

Kimi K3.1 Pro

Moonshot AI

Moonshot AI's flagship successor to Kimi K3, a 2.8-trillion-parameter mixture-of-experts model (104B active) for long-horizon coding and agentic work, with image and video input and a 1M-token context.

LP 1154#9League debut Oct 24 · the day after release

Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.

Flint

Recursive

Recursive's first model, a 40B mixture-of-experts coding model (2.5B active) built to run locally, at over 125 tokens per second on a MacBook, with a 262K-token context.

LP 576#29League debut Oct 22 · the day after release

Recursive Flint is here: our first model, built with the Eureka Machine. Fast, local coding. Open weights available on Hugging Face.

MiniMax M3.1 Flashmax

MiniMax

MiniMax's low-cost 428B mixture-of-experts model (23B active) for coding, agentic work and image/video input, with a 1M-token context.

LP 903#20League debut Oct 20 · the day after release

Welcome back. Space Bunny. 🚀 MiniMax-M3.1-Flash is coming.

Mercury 3high

Inception

Inception's diffusion language model, generating over 1,000 tokens per second, with a 1M-token context.

LP 890#22League debut Oct 21 · 2 days after release

Mercury 3 is live: the most capable diffusion LLM released to date. It outperforms GPT-6 Luna and Claude Haiku 5.5 across a range of benchmarks while running at over 1,000 tokens per second.

Beam 1.0max

Reflection

Reflection's open 501B mixture-of-experts model (23B active) for coding and agentic work.

LP 969#18League debut Oct 17 · the day after release

Beam is now available across platforms, with open weights under the Apache 2.0 license - in collaboration with an ecosystem of 40+ partners.

Nemotron 3.5 Super VLxhigh

NVIDIA

NVIDIA's open 120B (12B active) vision-language model for agents, with a 1M-token context.

LP 595#28League debut Oct 17 · 2 days after release

Meet NVIDIA Nemotron 3.5 Super VL Our new open model combines reasoning and tool use with text, image, document, and video understanding for long-running AI agents.

Cat Jammax

Undisclosed

A fast stealth model for agentic coding from an undisclosed developer.

LP 799#24RetiredLeague debut Oct 15 · 2 days after release

Cat Jam (stealth model) is now free in OpenCode - very fast - built for agentic coding - zero data retention

Muse Wonder 1.0max

Meta

Meta's frontier model for coding, long-running agents and advanced computer use.

LP 1046#13League debut Oct 13 · the day after release

Muse Wonder Full 1.0max

Meta

Meta's most advanced model, available only through a Muse Code subscription.

LP 1150#10League debut Oct 13 · the day after release

We’re excited to introduce the Muse Wonder series with Muse Wonder and Muse Wonder Full, our most capable models yet.