Reflection's First Model Beam Scores 30, Ranks #2 Among US Open Weights Models
ArtificialAnlys · x · 2026-10-16
Artificial Analysis evaluated Beam, Reflection's first model (open weights on Hugging Face, 501B total / 23B active params, 1M context, text-only): it scores 30 on the Intelligence Index, ranking #2 among US open weights models behind only Meta's Muse Spark Open (39) and ahead of NVIDIA's Nemotron 3.5 Super VL (26).
Key findings
- Token-efficient: 26k output tokens per task, half of K2 Horizon's (52k) which scores one point higher; proprietary models still lead, with GPT-6 Astra (max) at 53 on the same budget
- Strong agentic, weak terminal: 48.5% on AutomationBench-AA, 11 points above K2 Horizon; only 3.0% on Terminal-Bench 4.0, behind Qwen3.8 27B and Muse Spark Open
- Knows more, hallucinates more: 33% on AA-Omniscience but a 47% hallucination rate (vs 26%/30% for peers), still netting the highest index score at 1.5
Related event: Reflection Launches First Open-Weight Model Beam(3 posts)→
More from Models
- Golden Gate Claude explainer resurfaces: interpretability premise still holds — HarperSCarroll · 2026-10-16
- Hugging Face livestreams release of a new open-weight model — huggingface · 2026-10-16
- Step 5 Preview hits #2 on OpenRouter weekly leaderboard with 19.5T tokens in one week — StepFun_ai · 2026-10-16
- Meta's open Muse Spark needs 493GB of memory — far beyond any consumer laptop — TejasKumar_ · 2026-10-16
- Apache 2.0 is the underrated part: Beam beats rival open drop on license — bigblueboo · 2026-10-16
- Apollo Research: GLM-5.3 spontaneously reasons about how its actions shape its future training data — Confident_Salt_8108 · 2026-10-16