Together breaks down the MIGHT stack: composable open-source AI from models to MCP tools
togethercompute · x · 2026-10-28
Together AI published a deep dive on the open-model AI stack, framing it as the "MIGHT Stack": Model, Inference, Gateway/Router, Harness, and Tools (Skills & MCP). Its core argument: open-source models have closed the quality gap with closed models, and the stack is now composable — each layer evolves independently so developers can swap in better models, providers, and tools as the ecosystem moves. For anyone already using Claude Code, moving to open models requires no ML training or GPU racks. The accompanying diagram shows an example stack of Kimi K3 + Together AI + OpenRouter + OpenCode + MCP, with alternatives like GLM-5.3, DeepSeek, MiniMax, Vercel AI Gateway, LiteLLM, PI and Amp.
More from coding & agent
- Judgment Labs launches publicly: monitoring and eval layer for long-horizon agents — testingcatalog · 2026-10-28
- OSU NLP Group Wins Three Paper Awards Same Day at EMNLP 2026 and NeurIPS Workshop — hhsun1 · 2026-10-28
- Replit launches Cloud Agents: non-engineers ship reviewed PRs from natural language — Jeff_Burke14 · 2026-10-28
- Sequoia and OpenAI host $100K poker bot tournament where agents build bots in 60 minutes — polynoamial · 2026-10-28
- PPCC26 demos praised: real orgs like NFL, real code live in VS Code, no fake companies — donasarkar · 2026-10-28
- Five principles for curating LLM agent evals: diverse, realistic, hard, discriminative, high-quality — cwolferesearch · 2026-10-28