X Discussion: OpenAI's Tibo Calls Comparing GPT-6.1 Terra to Claude Opus 5.5 "Pretty Wild," Saying Terra Is Not Meant for the Big Leagues

Core Background & Industry Pain Points

When picking a model, developers often compare a cheaper, faster tier with a flagship as if they were aimed at the same job. James Malsawm (@JamesMalsawm) asked at 09:30 UTC on November 7, "Do you guys prefer Opus 5.5 or Gpt 6.1 Terra?"; at capture that question had about 197 likes and 87 replies. Tibo (@thsottiaux), who lists himself as working on Codex and ChatGPT at OpenAI, replied on X at 05:40 UTC on November 8 that comparing GPT 6.1 Terra to Opus 5.5 seemed "pretty wild" to him: Terra is "a great model" but "not meant to compete in the big leagues," and "we've got other models for that." At capture the reply had about 1,020 likes, 200 replies, 23 reposts, 18 quotes and about 74k impressions. This is his personal comment on X, not an official OpenAI announcement.

Architecture Highlights & Underlying Mechanisms

The discussion is about product positioning, not a new architecture. OpenAI's introduction on October 26, 2026 describes GPT-6.1 Terra as a fast, efficient model for detailed visual reasoning and precise computer use, with standard API prices of $1 per million input tokens and $5 per million output tokens, and it says that for the most complex coding tasks, GPT-6.1 Sol and GPT-6 Astra remain the better choice. Anthropic positions Claude Opus 5.5 as a hybrid reasoning model built for serious coding and AI agents, with API prices of $4 per million input tokens and $20 per million output tokens. Tibo did not name which models he meant by "the big leagues" in this reply, and he did not publish a new evaluation, release date or roadmap.

Benchmarks & Measured Results

This discussion did not add a new third-party benchmark. OpenAI's introduction (vendor figures) says that on Agents' Last Exam at medium reasoning effort, GPT-6.1 Terra scores 15 percentage points higher than GPT-5.6 Terra at about one-third the cost; at max effort, Terra is over three times as token efficient as Claude Opus 5.5 and one-fourth its price per token, resulting in about 9x lower cost per task; on Terminal-Bench 4.0 at max effort, Terra solves 50.3% of tasks, more than double GPT-5.6 Terra's published 21.5%. Anthropic's introduction (vendor figures) lists Opus 5.5 at 66.4% on Terminal-Bench 4.0 (xhigh). The two sets of numbers come from different vendors and setups, so they are not a conclusion of this X thread, and Tibo did not cite them.

Developer Hands-on Guide

For high-volume computer use, forms and visual understanding, the official model ID is gpt-6.1-terra, priced at $1 / $5 per million tokens. For long-horizon coding or agents, Anthropic's API ID is claude-opus-5-5, priced at $4 / $20 per million tokens. Tibo's point was only that Terra should not be treated as the model meant to go head-to-head with Opus 5.5, and that OpenAI has other models for that tier; he did not name them. Reading the official introductions is safer than treating one reply as a ranking. This article summarizes an X discussion; OpenAI has not launched a new model or an official ranking here.