GPT-6.1 Terra
OpenAI·
LLMsflagshipcloud
Overview
Mid tier of the GPT-6.1 family, released four weeks after GPT-6.1 Sol and the first Terra model of the GPT-6 generation. Priced at $1 / $5 per million input/output tokens ($2 / $7.50 above 272K input tokens), cached input $0.05. Rolling out in ChatGPT Work and Codex for paid plans and in the API as gpt-6.1-terra. 1.05M-token context.
Capabilities and innovations
Agentic codingComputer useVisual reasoningTool use / function callingReasoning effort control1.05M token context windowAPI: gpt-6.1-terraSol / Terra / Luna model family
Benchmarks
| Benchmark | Score | Source |
|---|---|---|
| Humanity’s Last Exam (no tools) | 46.2% | third party HLE 46.2% no-tools, measured by Artificial Analysis at max effort, "GPT-6.1 Terra (Max)" (accessed 2026-10-30). |
| Terminal-Bench 4.0 | 49.49% | independent Terminal-Bench 4.0 49.49% (± 1.82), measured by Vals AI via OpenAI's API at max effort, rank 11 of 58 (accessed 2026-10-30). Artificial Analysis measures 55.6 in its own run; the gap exceeds 3 points, so the Vals run is primary, as for GPT-6.1 Sol. 4.0 is its own generation and not comparable with the 2.x or 3.0 fields. |
| MMMU-Pro | 84.7% | third party MMMU-Pro 84.7%, measured by Artificial Analysis at max effort (accessed 2026-10-30). |
- GPQA Diamond: independent evaluation pending — OpenAI's launch post was not reachable from this environment. Artificial Analysis Intelligence Index v4.3 no longer runs GPQA Diamond (value null on the model page, checked 2026-10-30).
- SWE-bench Pro: independent evaluation pending — OpenAI reported DeepSWE v1.1 instead for GPT-6 Sol and GPT-6.1 Sol; the Terra launch table could not be read from this environment (checked 2026-10-30).
- MMLU-Pro: independent evaluation pending
- Terminal-Bench 2.x: independent evaluation pending — Vals AI and Artificial Analysis measure Terminal-Bench 4.0 only (recorded in terminal_bench_4); AA lists no 2.1 run for Terra.
- DeepSWE v1.1: independent evaluation pending — No v1.1 figure read at source: openai.com is not reachable from this environment (checked 2026-10-30).
- ARC-AGI-2: independent evaluation pending — No ARC Prize results page for GPT-6.1 Terra yet (arcprize.org/results/openai-gpt-6-1-terra returns 404, checked 2026-10-30).
API pricing
$1 per million input tokens, $5 per million output tokens (USD, cheapest OpenRouter endpoint, checked 2026-10-30)
Reliability
- omniscience_index
- 29.6 ()independent
Links
More from OpenAI
Data curated by AI Model Timeline. See the methodology for admission criteria, benchmark eras and source priority.