GPT-6.1 Terra benchmarked: half the price, big agentic gains
- Source
- x.com
- Date
OpenAI has released GPT-6.1 Terra, pushing the cost efficiency frontier. It is the first Terra-class model since GPT-5.6, and is more intelligent for less than a third of the cost Pricing per token is half or less that of GPT-5.6 Terra - at $1/$5 per million input/output tokens vs $2/$12 for the previous model. Cache write has also halved, now at $1.25 per million tokens. Cached input is one quarter at $0.05 vs $0.20 per million tokens. Key takeaways: ➤ Pareto-optimal for Intelligence vs Cost at low and medium effort: At max effort, GPT-6.1 Terra gains 6 points in the Artificial Analysis Intelligence Index over GPT-5.6 Terra, at less than a third of the cost per task ($0.40 vs $1.40). At lower effort levels the gain is larger, between 7 and 13 points. The low and medium effort levels are Pareto-optimal for cost efficiency, while the other effort levels sit behind the frontier. At max effort, the model scores 4 points below GPT-6.1 Sol and 10 points above GPT-6 Luna in the Intelligence Index. ➤ Pushes the Coding Agent Index cost efficiency frontier: With a score of 48 in the newly released Artificial Analysis Coding Agent Index v2.0, the model is 6 points below GPT-6.1 Sol (54)…
Context
Artificial Analysis, which publishes its own model indices, found that OpenAI's GPT-6.1 Terra scores higher than the previous Terra model while costing much less to run. At its maximum effort setting, GPT-6.1 Terra gains 6 points over GPT-5.6 Terra on the Artificial Analysis Intelligence Index, and a task costs $0.40 instead of $1.40. List prices also fell: $1 per million input and $5 per million output tokens, against $2 and $12 before, with cached input dropping from $0.20 to $0.05 per million tokens, according to the Artificial Analysis post.
The gains are largest at the cheaper settings. Effort settings let a user trade cost for capability, and at lower settings the gain over GPT-5.6 Terra grows to between 7 and 13 points. Artificial Analysis calls the low and medium settings Pareto-optimal, meaning no other option it measured scored higher on its index for the same cost or less. The other settings sit behind that line on this index, so they buy capability less efficiently. Even at maximum effort, Terra stays 4 points below GPT-6.1 Sol and 10 points above GPT-6 Luna on the index.
The largest single change is in how often the model guesses. On AA-Omniscience, a private Artificial Analysis uses to measure knowledge and , GPT-6.1 Terra guessed 38% of the time when it did not know an answer, down from 88% for GPT-5.6 Terra. Its accuracy on the same benchmark rose 2 points, so the lower guess rate did not come from answering fewer questions correctly.
-style tests show a similar pattern. GPT-6.1 Terra scores 48 on the new Artificial Analysis Coding Agent Index v2.0, against 54 for GPT-6.1 Sol and 31 for GPT-6 Luna, and it gained about 20 points on Terminal-Bench 4.0, a benchmark of agentic work in a command-line terminal, where Artificial Analysis says it now rivals GPT-6.1 Sol at maximum effort. Most other evaluations improved by smaller amounts, such as 5 points on GDP.pdf. These are one benchmarker's results on its own indices, including a private one, but they support OpenAI's framing in its launch post that Terra does more at a lower price. Its low effort setting sits on Artificial Analysis's cost-efficiency frontier for both general intelligence and coding agents, while Sol still leads on overall scores.
- token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
- benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
- hallucination — When a model states something false with full confidence — inventing facts, citations, or APIs that don't exist.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Checking sign-in…
Loading comments…