Neon runs 2x faster than Carbon at 110 tok/s, raising doubts about separate pretrain claim
teortaxesTex · x · 2026-11-19
teortaxesTex continues analyzing the Argon/Carbon/Neon family, arguing the vendor's MRL framing is a handwave implementable in many ways.
Benchmark screenshots show Neon at #73/95 with 56.51s first-token latency, 110 tok/s output and 21.5 min per Intelligence Index task, vs Carbon at #91/95 with 72.86s, 56 tok/s and 36.6 min — a classic "Flash-tier" profile that's 2x faster, more verbose and marginally cheaper per token, yet the vendor doesn't bill Neon as a separate pretrain run.
Related event: Argon/Carbon/Neon Models Suspected of Distilling from Same Source(2 posts)→
More from Models
- Gemini 4 breakdown: Neon is 2x faster than Carbon but hallucinates more — teortaxesTex · 2026-11-19
- Gemini 4 Neon (High) beats Hydrogen and Carbon on AI SEO advice in hands-on test — gaganghotra_ · 2026-11-19
- Agents spent years hillclimbmaxxing while the human at the keyboard became an afterthought — tokenbender · 2026-11-19
- Speculation: Argon/Carbon/Neon may be nested distillations, with 'adapted' dodging 'post-trained' — teortaxesTex · 2026-11-19
- 6.2-Astra reportedly crushes contest benchmarks, finding a novel AWTF26 solution in 2 hours — i_dg23 · 2026-11-19
- KVARK Flash: a proxy that caps DeepSeek's thinking per request, 30-50% fewer tokens — Ill_Contribution_717 · 2026-11-19