Grok 4.5 Ties with GPT-5.6 in Benchmark
scaling01 · x · 2026-07-10
The post shares a KernelBench-Hard update, noting that Grok 4.5 high and GPT-5.6 Sol xhigh achieved roughly tied overall scores on the RTX PRO 6000.
It further breaks down performance across multiple sub-categories, including FP8, paged attention, W4A16, and KDA, highlighting the strengths and weaknesses of each model across different operators and Triton/cuBLAS pathways.
More from Models
- Rumor claims Kimi faked performance by serving Claude; DeepSeek new model surprises in evals — realsohamparekh · 2026-09-11
- GPT-5.6 writes well but is instantly forgettable, user complains — BasedRaddka · 2026-09-11
- Opus Refuses Protein Research Codebase Over 'Safety' Concerns, Dev Considers Rolling His Own — josephdviviano · 2026-09-11
- User Hails Unconfirmed 'DeepSeek 4.1 Flash' as an Inflection Point in LLMs — himanshustwts · 2026-09-11
- Terminal Bench v4: GLM-5.3 Leads at 41.9%, Kimi-K3 Underwhelms at 12.6% — Ok_Warning2146 · 2026-09-11
- GPT-6 Astra beats Factorio with enemies in 44 in-game hours at ~$4,500 API cost — liminal_bardo · 2026-09-11