Thinking Machines' Inkling 2 Small (594B MoE, 1M context) lands on Ollama
ollama · x · 2026-11-13
Inkling 2 Small from Thinking Machines is now on Ollama Cloud with zero data retention (ollama run inkling-2-small:cloud). The 594B mixture-of-experts model has a 1M context window and targets visual interpretation/creation, agentic knowledge work, and finance/quantitative reasoning (OCR, grounding, MCP tool calling, filings analysis).
Benchmark charts vs GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, and Kimi K3 show:
- Visual reasoning is its edge: MMMU Pro 81.6%, PerceptionBench 64.7%, ScreenSpot Pro 83.8/92.0%, GDP.pdf 25.2% — mostly first place
- Coding near the top: SWEBench Pro v2 86.3/97.7%, Vibe Code Bench 81.6%, though behind DeepSeek-V4.1-Flash (89.7%, 87.7%)
- Strong enterprise knowledge work: best on JobBench 70.5%, EnterpriseBench-CoreCraft 71.8%, DAYJOB (Healthcare) 6.4%
Related event: Thinking Machines' Inkling 2 Small lands on Ollama with 1M context(2 posts)→
More from coding & agent
- New research tackles the agent experience problem: knowledge trapped in weights — khademinori · 2026-11-13
- YC F26's Acyclic Labs open-sources infra to run 10,000-agent swarms in one session — ycombinator · 2026-11-13
- Floci: free local AWS emulator with 31.6k GitHub stars for faster integration tests — JeremyCMorgan · 2026-11-13
- Open-source REM tool reverse-engineers competitors' ads, SEO and social playbook from one URL — moeinteractive · 2026-11-13
- Codex agent shrinks a 27M-param transformer to 11.46M with equal loss via weight sharing — yoshiyama_akira · 2026-11-13
- Google open-sources Harness²: Agent = Model + Harness, recursive agent harnessing framework — 2C_ornot2C · 2026-11-13