Claude Haiku 5.5 (Xhigh) vs GPT-6.1 Terra (High)
Which is better for the money: Claude Haiku 5.5 (Xhigh) by Anthropic or GPT-6.1 Terra (High) by OpenAI? Intelligence, total benchmark cost, and end-to-end time from Artificial Analysis & other sources, snapshot 2026-10-27.
Trade-off
Verdict
Depends on your budget
GPT-6.1 Terra (High) scores 2.7 points higher (43.9 vs 41.2) but costs 1.5× as much to run the benchmark ($239 vs $158; both mid-range class). Pick GPT-6.1 Terra (High) when the extra intelligence is worth the premium, otherwise Claude Haiku 5.5 (Xhigh).
| Metric | Claude Haiku 5.5 (Xhigh) | GPT-6.1 Terra (High) | Difference |
|---|---|---|---|
| Intelligence (higher is better) | 41.2 | 43.9 | +6.6% GPT-6.1 Terra (High) is 6.6% more intelligent |
| Total AA cost (lower is better) | $157.58 | $238.53 | +51.4% GPT-6.1 Terra (High) is 51.4% more expensive |
| End-to-end time (lower is better) | 75.35s | 11.16s | −85.2% GPT-6.1 Terra (High) is 85.2% faster |
| Price per 1M tokens (in / out) | $0.10 / $0.50 | $1.00 / $5.00 | |
| Input modalities | Text, Image | Text, Image |
GPT-6.1 Terra (High) vs Claude Haiku 5.5 (Xhigh): 6.6% more intelligence, 51.4% more cost to run the AA benchmark, and 85.2% faster end-to-end.
Today's Pareto frontier
Against the frontier
- Claude Haiku 5.5 (Xhigh) is on today's frontier: nothing is both smarter and cheaper.
- GPT-6.1 Terra (High) is off the frontier: at the same or lower cost, MiMo-V2.6-Pro scores 46.30 for $207.
Claude Haiku 5.5 (Xhigh)GPT-6.1 Terra (High)Today's Pareto frontierEvery other model
Matchup
Compare something else
Agents:
Accept: text/markdown or ?format=md returns this page as Markdown. JSON: /api/v1/compare/claude-haiku-5-5-xhigh/gpt-6-1-terra-high · Markdown: /api/v1/compare/claude-haiku-5-5-xhigh/gpt-6-1-terra-high?format=md · llms.txt · agent skill