Uncensored 26.4B model runs on a 12GB Mac at 85 tok/s with 8 parallel agents via Locals
SumitGup · x · 2026-11-23
- Uncensored Inkling 2 Mini (26.4B) has been adapted for M4 Apple Silicon and runs locally on a 12GB laptop: refusals dropped from 100% to 0% in testing while keeping 98% of MMLU accuracy, with three GGUF quants released (8.67–16.53 GB).
- Using Locals, an open-source local inference server built for agents (660 GitHub stars), the M4 Mac reaches 85 tok/s with 8 sub-agents, 71 tok/s rewriting files via tools, and 61 tok/s writing new files. Locals supports tool calls, long conversations, and many concurrent agents, working with OpenCode, OpenHands, Cline, Aider, Claude Code, and any OpenAI/Anthropic/Ollama app, claiming up to 3x faster multi-agent inference.
More from coding & agent
- Only-Knowing-Code Developers at Risk as the Agentic Software Factory Reshapes Engineering — Pavan_Belagatti · 2026-11-23
- MCP Tool Gateways Should Fail Loudly: A Plausible Answer From a Missing Dependency Is the Real Bug — Emergency-Tutor-7 · 2026-11-23
- Codex Can't Route PCB Traces: Devs Still Lack a Workable AI Tool for Hardware's Last Mile — goldvenetianmask · 2026-11-23
- spaCy creator Honnibal shares a cheap-LLM ensemble recipe for building NER data with coding agents — honnibal · 2026-11-23
- 398M tokens, 98% cached: how one dev ran 12.5h of agent game dev on 3% of a $200 Claude plan — Yamapama · 2026-11-23
- Harvard paper: higher SWE-Bench scores didn't translate into more merged pull requests — rohanpaul_ai · 2026-11-23