Apple's AFM3 still has no public benchmark scores, only human preference evals
Recoil42 · reddit · 2026-09-18
Apple's third-generation foundation models (AFM3 Core/Advanced) launched without any standard benchmark results — Apple's own announcements only cite human preference evals over AFM 2025. Redditors are still waiting to see how AFM3 stacks up against new Qwen and Gemma models.
More from Models
- xAI insider says Grok 4.7 is still 'in the oven' after user asks where it is — ChrisUniverse · 2026-09-18
- Jev Ditches Autoregression: A Model That Only Outputs Structured Decisions — karminski3 · 2026-09-18
- MLX Community Makes Qwen 3.8 Flash Nearly 2x Faster on Apple Silicon, License Blocks Launch — gajesh · 2026-09-18
- PrismML Bets Its Tiny LLM Can Change How We All Use AI — TechCrunch AI · 2026-09-18
- Hand-written paper with 3 copied AI sentences — Pangram flagged exactly those three — johnowhitaker · 2026-09-18
- Dev reimplements DeepSeek v4.1 Flash, runs 1M context at 1-10 tok/s on a single RTX 4090 — _xjdr · 2026-09-18