GPT-6.1 Terra impresses with speed and price, splits opinions in real-world tests

OpenAI released GPT-6.1 Terra (with the max tier focused on computer use), priced at $1/M input and $5/M output—half of GPT-5.6—claiming it is "9x cheaper per task." Community benchmarks poured in after launch; the consensus is that its speed and cost advantages are clear, while its capability ceiling falls short of the Opus tier, making "cheaper, not better" the prevailing verdict.

Confirmed

• Pricing: $1/M input, $5/M output, half of GPT-5.6; @aigenerator2077 broke down the "9x cheaper" claim, attributing it to factors like token efficiency but found the math doesn't add up (relayed via m11).
• Multiple independent tests show cost and speed advantages: on the same rainbow gummy bear task, Terra took $4.15 and 25 minutes vs Opus 5.5 at $13.16 and 40 minutes (m3); @VraserX reported the same coding task ran 3x faster than Opus 5.5 at roughly one-ninth the cost, with performance close to Opus (m5); @robocurve's hands-on test said Terra beat Opus 5.5 at controlling a real robot, a 3.7x improvement over three months ago (relayed via m1).
• Artificial Analysis's "intelligence index vs cost per task" scatter plot (covering 78 models) shows GPT-6.1 Terra medium/high on the Pareto frontier in the most attractive quadrant, with the frontier almost entirely occupied by OpenAI models, from entry-level at $0.006/task up to the $1+ range (m4, m6).
• Speed data: Terra far ahead of Sol (91 vs 61-81 tokens/s, via @haider1 relaying m7); @chongdashu's coding tests found capabilities clearly below the Sol/Astra/Opus tiers but blazingly fast—tasks finished before switching sessions was even considered (m9).
• Negative takes: Reddit user bofors14 said Terra is weak across the board at coding; as a sub-agent (mid/low tier) it only added cost with no benefit versus luna max, and it's not ideal as the main agent either (m8); in Towards AI's ToneBench writing evaluation (10 YouTube script tasks × 5 drafts, blind-judged by 4 LLMs), Terra ranked #23, concluding "cheaper, not better" (m10).

Unconfirmed

• The cost and performance figures are all individual or third-party self-built evaluations, not official benchmarks, and highly scenario-dependent; @IamNotGorbachev noted polarized community opinions ("useless," "terrible," and "powerful" coexist), with outcomes heavily dependent on use case.
• @robocurve's robot-control data is secondhand; original details cannot be verified.

Why it matters

Terra represents OpenAI's strategy of capturing the enterprise agent/computer use market with a cheap, fast model: on the intelligence-cost curve, OpenAI nearly monopolizes the Pareto frontier, directly undercutting competitors' value-for-money narratives. For developers, though, the trade-off between "good enough and cheap" versus "top-tier capability" is theirs to make—it suits high-frequency, cost-sensitive tasks but doesn't replace the Opus tier for complex work.

2026-10-28 ~ 2026-10-29 · 11 related posts

• [source] GPT-6.1 Terra nearly matches Claude Opus 5.5 at 3x speed and 1/9th cost — VraserX · 2026-10-28
• ToneBench eval: GPT-6.1 Terra ranks #23 at writing — cheaper, not better, than GPT-5.6 Terra — OnlyProggingForFun · 2026-10-28
• GPT-6.1 Terra beats Opus 5.5 at real robot control, 3.7x better than GPT-5.6 from 3 months ago — ycombinator · 2026-10-28
• [source] First Hands-On: GPT 6.1 Terra Is Blazing Fast but Below Opus Tier — chongdashu · 2026-10-28
• GPT-6.1 Terra priced at $1/$5 per M tokens, but the '9x cheaper per task' math doesn't hold — 2C_ornot2C · 2026-10-28
• GPT-6.1 Terra beats Opus 5.5 on same prompt at $4.15 vs $13.16 and 25 vs 40 minutes — dkundel · 2026-10-28
• GPT-6.1 Terra earns its spot: medium and high tiers land on the cost-intelligence Pareto frontier — Balance- · 2026-10-28
• GPT-6.1 Terra vs Sol: enterprises want cheap-good-enough models, says analysis — haider1 · 2026-10-29
• Chart shows OpenAI dominating the intelligence-per-cost Pareto frontier across 78 models — Lucky_Creme_5208 · 2026-10-29
• Redditor: terra is 'absolutely useless' for coding in both subagent and main agent roles — bofors14 · 2026-10-29
• [source] OpenAI Terra 6.1 in practice: cheap model plus tiered follow-up beats what benchmarks miss — IamNotGorbachev · 2026-10-29