GPT-5.6 Terra (Non-reasoning) vs Beam 1.0
Which is better for the money: GPT-5.6 Terra (Non-reasoning) by OpenAI 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
Beam 1.0 scores 9.5 points higher (30.3 vs 20.8) but costs 1.2× as much to run the benchmark ($319 vs $256; both mid-range class). Pick Beam 1.0 when the extra intelligence is worth the premium, otherwise GPT-5.6 Terra (Non-reasoning).
| Metric | GPT-5.6 Terra (Non-reasoning) | Beam 1.0 | Difference |
|---|---|---|---|
| Intelligence (higher is better) | 20.8 | 30.3 | +45.7% Beam 1.0 is 45.7% more intelligent |
| Total AA cost (lower is better) | $255.77 | $318.73 | +24.6% Beam 1.0 is 24.6% more expensive |
| End-to-end time (lower is better) | 5.09s | 8.02s | +57.6% Beam 1.0 is 57.6% slower |
| Price per 1M tokens (in / out) | $2.00 / $12.0 | $0.40 / $1.20 | |
| Input modalities | Text, Image | Text |
Beam 1.0 vs GPT-5.6 Terra (Non-reasoning): 45.7% more intelligence, 24.6% more cost to run the AA benchmark, and 57.6% slower end-to-end.
Today's Pareto frontier
Against the frontier
- GPT-5.6 Terra (Non-reasoning) is off the frontier: at the same or lower cost, MiMo-V2.6-Pro scores 46.30 for $207.
- Beam 1.0 is off the frontier: at the same or lower cost, MiMo-V2.6-Pro scores 46.30 for $207.
GPT-5.6 Terra (Non-reasoning)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/gpt-5-6-terra-non-reasoning/reflection-beam · Markdown: /api/v1/compare/gpt-5-6-terra-non-reasoning/reflection-beam?format=md · llms.txt · agent skill