Qwen3.8 27B (Xhigh) vs Beam 1.0
Which is better for the money: Qwen3.8 27B (Xhigh) by Alibaba or Beam 1.0 by Reflection? Intelligence, total benchmark cost, and end-to-end time from Artificial Analysis & other sources, snapshot 2026-10-24.
Trade-off
Verdict
Depends on your budget
Qwen3.8 27B (Xhigh) scores 3.4 points higher (33.7 vs 30.3) but costs 4.2× as much to run the benchmark ($1336 vs $319; premium vs mid-range cost classes). They are not alternatives at one budget: pick Qwen3.8 27B (Xhigh) when you need intelligence above 30.3, and Beam 1.0 when 30.3 is enough.
| Metric | Qwen3.8 27B (Xhigh) | Beam 1.0 | Difference |
|---|---|---|---|
| Intelligence (higher is better) | 33.7 | 30.3 | −10.1% Beam 1.0 is 10.1% less intelligent |
| Total AA cost (lower is better) | $1,336.09 | $318.73 | −76.1% Beam 1.0 is 76.1% cheaper |
| End-to-end time (lower is better) | 60.10s | 8.02s | −86.7% Beam 1.0 is 86.7% faster |
| Price per 1M tokens (in / out) | $0.50 / $3.00 | $0.40 / $1.20 | |
| Input modalities | Text, Image, Video | Text |
Beam 1.0 vs Qwen3.8 27B (Xhigh): 10.1% less intelligence, 76.1% less cost to run the AA benchmark, and 86.7% faster end-to-end.
Today's Pareto frontier
Against the frontier
- Qwen3.8 27B (Xhigh) is off the frontier: at the same or lower cost, GPT-6.1 Sol (Max) scores 51.80 for $1082.
- Beam 1.0 is off the frontier: at the same or lower cost, MiMo-V2.6-Pro scores 46.30 for $207.
Qwen3.8 27B (Xhigh)Beam 1.0Today's Pareto frontierOpen-weights frontierEvery other model
Matchup
Compare something else
Agents:
Accept: text/markdown or ?format=md returns this page as Markdown. JSON: /api/v1/compare/qwen3-8-27b/reflection-beam · Markdown: /api/v1/compare/qwen3-8-27b/reflection-beam?format=md · llms.txt · agent skill