Model results · OpenAI

GPT-6.1 Terra

Every published VulcanBench measurement of GPT-6.1 Terra, newest first: combined score, Code quality, runtime, tokens and API-equivalent cost at every effort level, graded by deterministic hidden tests on real tasks. Measured through the Codex CLI on a ChatGPT subscription. GPT-6.1 Terra is a newer model than GPT-5.6 Terra, which has its own page.

Suitev4

GPT-6.1 Terra across every effort level

In Codex across five effort levels: combined score 74.95 at Low, 80.10 at Medium, 79.19 at High, 83.68 at Extra-high and 90.00 at Max. Tasks passed 13, 12, 15, 16 and 23 of 23, with every hidden test passed at Max; Code quality 66.55 to 76.59; 2.8 to 6.9 minutes per task. Every cell is judged on 23 runs.

October 30, 2026 · 23 tasks · 5 effort levels · 115 runs of GPT-6.1 Terra · VulcanBench Frontier v4 · Code quality protocol v3.29

→

Suitev4

Beside Claude Opus 5.5, GPT-6.1 Sol and GPT-6 Astra

At Max GPT-6.1 Terra scores 90.00, between Claude Opus 5.5 (90.22) and GPT-6 Astra (89.30), and is the cheapest and fastest of the four at every level: $0.36 and 6.9 minutes a task at Max. On the Frontier v4 board its Max level ranks 16th of 80 columns.

October 30, 2026 · comparison card · VulcanBench Frontier v4

→

Suitev4

Cost and tokens across effort levels

$0.11 to $0.36 per task at list API rates and 0.77M to 2.38M raw tokens per task across the five effort levels; the whole sweep prices at $22.49 for 115 runs, $0.20 per task.

October 30, 2026 · API-equivalent estimates at list rates, solver inference only · VulcanBench Frontier v4

→

Tracesv4

Redacted agent traces: GPT-6.1 Terra on Frontier v4

The trace repository for the sweep in the report above, published separately on GitHub. Structure only: every agent message, shell command and file change with its timing, and a closing summary per run, without the task content.

GitHub · vulcanbench-gpt61-terra-traces

→

How would GPT-6.1 Terra do on your stack?

Public results answer "which model is stronger on this suite." Whether GPT-6.1 Terra belongs in your routing table depends on your languages, your codebases, and your task mix, that gets measured, not guessed. Grading rules are in the methodology.

Optimize your model routing All reports