DeepSeek V4.1 semi-consistently identifies itself in chat, seen as a sign of stronger RL
teortaxesTex · x · 2026-09-08
teortaxesTex observes that DeepSeek V4.1 is the only model that semi-consistently identifies as DeepSeek in extended self-exploration (sometimes drifting to Gemini or "nothing in particular"), suggesting stronger RL. The model's novel, curious manner of exploring its own identity is also noted.
More from Models
- Blender head-to-head: same prompt, 12 seconds, and one frontier model is in a different class — ZeroStateReflex · 2026-09-09
- GLM-5.3-Flash tops agentic tool-call leaderboard at 78%, priced at just $0.50/M output tokens — shensi · 2026-09-09
- Artificial Analysis Updated Benchmarks Twice in 4 Days for Astra — py-net · 2026-09-09
- Gary Marcus: it was the harness, not the model — hidden risk of opaque scaffolding — GaryMarcus · 2026-09-09
- Inception ships Mercury 2.5: most capable diffusion LLM at 1,107 tokens/sec, 260K context — StefanoErmon · 2026-09-09
- Bittensor miners score 42.3 on sports vision eval while GPT-6 Astra Ultra scores 0 — markjeffrey · 2026-09-09