17B-active Inkling 2 Small scores 31.87 on PostTrainBench, beating Kimi K3 and GLM 5.2
karinanguyen · x · 2026-11-06
thoughtfullab partnered with Thinking Machines to evaluate Inkling 2 Small, an open model claimed to show frontier-level performance at autonomously improving other models.
On PostTrainBench — where each agent gets four small base LLMs, one H100 GPU and 10 hours to post-train them — Inkling 2 Small scored 31.87 with only 17B active parameters, ahead of Kimi K3 (31.48) and GLM 5.2 (30.40), on par with GPT-5.6 (Sol) and Opus 4.8 (31), though well behind Fable 5.1 Max (44.6) and Opus 5.5 Max (44.8). Baselines: base models zero-shot score just 8.0, official instruct models 48.4.
More from Models
- NaceAI open-sources Drex decision models, hitting SOTA 67.52 on Decision Index — nischay_twt · 2026-11-06
- Users say Claude Opus 4.8 is 'so stubborn even its own classifiers can't stop it' — repligate · 2026-11-06
- Miles Brundage: Inkling 2 Small is good — Miles_Brundage · 2026-11-06
- Users claim Claude performs better with an empty system prompt than Anthropic's official one — repligate · 2026-11-06
- Palisade's Jeff Ladish: models better at everything but still not funny or persuasive — JeffLadish · 2026-11-06
- Palisade's Jeff Ladish calls to revive METR time-horizon estimates for new models — JeffLadish · 2026-11-06