OpenAdapter is an all-in-one affordable coding plan. Access the best open-source AI models — DeepSeek, Qwen, Mistral, Kimi K2.5, MiniMax, GLM-5, and more — in any editor. One plan starting at $9/mo.

Which AI models are included?

40+ open-source models across providers including DeepSeek, Qwen, Mistral, Kimi · Moonshot, MiniMax, GLM · Zhipu, Meta Llama, NousResearch, and more. New SOTA models are added on a rolling basis as they become available.

Which editors and tools are supported?

Cursor, VS Code (with Cline, Continue.dev, Roo Code), Windsurf, Claude Code, Aider, OpenCode, Zed, Void, PearAI, and more. Any editor with a custom base URL works.

Do I need to manage API keys?

No. OpenAdapter gives you one endpoint and one subscription. No juggling API keys from multiple providers.

Is there a free tier?

Yes — the Free plan includes 40+ open-source SOTA models with a strict daily quota, no credit card required.

MiMo-V2.6-Flash — the everyday omnimodal model, now with real agent chops — Changelog

MiMo-V2.6-Flash — the everyday omnimodal model, now with real agent chops

September 23, 2026

MiMo-V2.6-Flash is live. Xiaomi's efficiency checkpoint of the new V2.6 series: a 309B mixture-of-experts model with 15B active, a 1M-token context window, and text, image, video and audio input. MIT licensed. Call it as MiMo-V2.6-Flash.

Xiaomi's published scores, with the previous V2.5 Pro flagship for scale:

Benchmark	V2.6 Flash	V2.5 Pro
Terminal Bench 2.1	87.6	65.2
DeepSWE v1.1	67.9	19.0
Toolathlon-Verified	73.6	49.1
OSWorld-Verified	80.8	—
CyberGym	95.1	40.0
Agents' Last Exam	27.6	13.2

It costs 1.5 requests of quota per call and is available on every paid plan and the free tier, the same terms as MiMo-V2.5. It also joins the auto-pro pool as soon as it has enough traffic to be ranked.

Source: MiMo-V2.6-Flash model card.