26.4B Inkling 2 Mini matches 975B model in 128 days, ~17x compute efficiency gain
teortaxesTex · x · 2026-11-21
Thinking Machines Lab's Inkling 2 Mini (26.4B total / 3.4B active) reportedly surpasses a 975B model on Vals AI evals — a geometric-mean 17x compute efficiency gain (possibly 50x on pretraining) achieved in just 128 days.
Benchmark table highlights vs Gemma 4 31B, Muse Glimmer, Qwen3.8-27B, Nemotron 3.5 Super VL:
- SWEBench Verified (bash only) 79.2%, SWEBench Pro v2 Hard 77.9%, SWEBench Multilingual 73.7%
- DeepSWE v1.1 48.2%, FrontierCode v2.0 15.1%
- Agentic search: BrowseComp 79.0%, DeepSearchQA 85.3%, MCP Atlas 82.7%
- AA Index (v5.0) 20.1, tied with Qwen3.8-27B; Vals Index 46.5% far ahead
The Mini improves on the original Inkling in 13 of 18 shared benchmarks while being much smaller.
More from Models
- Waddle-1 robot-use agent scores 81 on unseen bimanual tasks, beating GPT-6.2 Astra at 66 — ycombinator · 2026-11-21
- alphaXiv Launches OpenResearch Intelligence Index, First Open Benchmark for Autoresearch — togethercompute · 2026-11-21
- Coding images of France with 6.2 Astra: a stress test for visual reasoning — prd_008 · 2026-11-21
- Hyperstition introduces judge-time scaling: a new scaling law for RL — eliebakouch · 2026-11-21
- Downdetector reports OpenAI outage starting 3:13 PM EST as user complaints pile up — VoidStateKate · 2026-11-21
- Inkling 2 Mini on Ollama: MLX build, one-command hooks into Claude Code and OpenCode — ollama · 2026-11-21