Towards AITowards AIToneBench

Model Comparison

Claude Opus 5.5 vs GPT-5.6 Terra

Claude Opus 5.5 (xhigh) leads GPT-5.6 Terra (ultra) by 711 Elo, and their 95% ranges sit 611 Elo apart. Even Claude Opus 5.5's lowest-rated configuration, Claude Opus 5.5 (low), ranks above GPT-5.6 Terra (ultra). Among 87 models, each ranked by its best configuration, Claude Opus 5.5 is #1 and GPT-5.6 Terra is #16. Cost per script: $1.00 for Claude Opus 5.5 (xhigh) and $0.275 for GPT-5.6 Terra (ultra).

Last updated , from board data generated at 17:52 UTC.

Claude Opus 5.5
2704 Writing Elo of Claude Opus 5.5 (xhigh), its best configuration#1 of 87 models · 85.8/100 Writing Score · $1.00 per script5 configurations on the board
GPT-5.6 Terra
1993 Writing Elo of GPT-5.6 Terra (ultra), its best configuration#16 of 87 models · 81.4/100 Writing Score · $0.275 per script2 configurations on the board

Picks and Verdict

PickClaude Opus 5.5

Claude Opus 5.5 (low)

Pick Claude Opus 5.5 (low) if you were weighing Terra on price, since $0.243 a script buys a draft 432 Elo above Terra's best in 1.5 minutes.

PickGPT-5.6 Terra

GPT-5.6 Terra (medium)

Pick GPT-5.6 Terra (medium) for a lot of drafts at $0.062 each, since ultra costs about 4.4 times as much for 71 more Elo and a lower Tone & Voice Match.

Claude Opus 5.5 (low) costs a little less than GPT-5.6 Terra (ultra), $0.243 against $0.275 a script, and that's the comparison that settles it for me. Opus at low ranks 432 Elo higher, with separate 95% ranges, and finishes in 1.5 minutes where Terra takes 8.0. Claude Opus 5.5 (xhigh) stretches that lead over Terra's best to 711 Elo, the #1 model of 87 against the #16 model. Terra's own ladder has one step, and it's a pricey one. Ultra adds 71 Elo over medium for about 4.4 times the cost, with ranges that overlap, and medium still scores higher on Tone & Voice Match, 78.6 against 76.9. If I ran Terra for volume, I'd keep it at medium. Opus has its own setting I'd skip on cost, max effort, at $2.58 for 5 fewer Elo and a range that overlaps xhigh's. Best against best, Terra holds up on substance, about two points back, and trails by about six on voice. For a script that has to sound like me, voice decides it.

Disclosure: Claude Opus 5.5 and GPT-5.6 Terra each come from the same lab as one of the judges (Claude Opus 5.5 and GPT-6.1 Sol). Neither is scored by its own lab alone (how the panel works). The judge analysis measures how much each judge favors its own lab.

How to Read the Writing Score, Writing Elo and 95% Ranges

Writing Score and Writing Elo. The Writing Score (0 to 100) is the weighted average of the nine metric scores: seven from the judges, Length Adherence from code and the Slop Score half from each. The Writing Elo is a rating from pairwise comparisons: on each script, each configuration's average across its drafts is compared with every other configuration's, and gaps within the draft-to-draft variation count as draws. Elo rewards winning often, so the two can put configurations in a different order.

Elo ranges. Each 95% range shows how far a configuration's Elo could move if the drafts had come out slightly differently. It comes from resampling every configuration's recorded draft scores on each reference script and ranking again. Comparing two separate ranges does not test the Elo gap between them: ranges far apart are strong evidence for the order, and overlapping ranges do not mean a tie. The reference scripts stay fixed, so the ranges say nothing about new scripts. The methodology page has the full description.

All Thinking Levels of Both Models

Every ranked configuration of Claude Opus 5.5 and GPT-5.6 Terra, scored on the same 10 scripts by the same four judges.

How to Read the Chart
  • Each row is one configuration: the model at one thinking level. The dot is its Writing Elo, a rating from pairwise comparisons of the scores on each script, and the bar is its 95% range: how far that Elo could move if the drafts had come out slightly differently.
  • Each bar describes one configuration. Overlap alone does not establish a tie or decide their order. Comparing these ranges does not directly test the difference between a pair.
  • The table under the chart lists the same configurations in the same order and adds the Writing Score (the 0 to 100 consensus of the judges and code), cost, time and configuration rank (the place among all configurations on the board).
Claude Opus 5.5GPT-5.6 Terra
ConfigurationWriting Elo95% RangeWriting ScoreCost / ScriptTime / ScriptConfiguration Rank
Claude Opus 5.5 (xhigh)27042648 to 275085.8$1.006.7 min#1
Claude Opus 5.5 (max effort)26992640 to 275585.9$2.5817.5 min#2
Claude Opus 5.5 (high)24712418 to 251884.6$0.3282.1 min#8
Claude Opus 5.5 (medium)24522395 to 250484.5$0.3101.9 min#9
Claude Opus 5.5 (low)24252359 to 248084.3$0.2431.5 min#10
GPT-5.6 Terra (ultra)19931952 to 203781.4$0.2758.0 min#32
GPT-5.6 Terra (medium)19221879 to 195780.7$0.0621.4 min#40

Cost is one script at the API list price of the route used, uncached and without judging. Time is the average for one accepted script, retries included.

Computed Findings Across Thinking Levels

Each finding is computed from the configurations above. The 95% ranges describe individual configurations; overlap alone establishes neither a tie nor an order for a pair.

Two Chosen Configurations Compared

Pick one configuration of each model to compare their metric scores, consistency across drafts and per-script scores. Each menu starts on the model's best configuration.

Scores by Metric

Claude Opus 5.5 (xhigh)GPT-5.6 Terra (ultra)Solid bar: higher score in that row

Claude Opus 5.5 (xhigh) scores higher on all nine metrics. Weighted by each metric's share of the Writing Score, Tone & Voice Match does the most to put Claude Opus 5.5 (xhigh) ahead, with a 5.6-point lead on that metric. The weight beside each metric is its share of the Writing Score.

Tone & Voice Match19% weight
82.5
76.9
Writing Craft & Clarity13% weight
83.2
76.8
Substance, Accuracy & Value15% weight
88.4
86.4
Continuity & Emotion14% weight
85.4
80.2
YouTube Best Practices12% weight
84.4
79.3
Hook Strength10% weight
82.9
81.5
Length Adherence8% weight
98.6
93.7
Slop Score5% weight
88.5
84.7
Visual Cue Quality4% weight
84.9
80.3

Consistency Across Drafts

Claude Opus 5.5 (xhigh)GPT-5.6 Terra (ultra)
Score spread across drafts (±)1.431.38
Lowest script score84.3 Script 1079.8 Script 4
Highest script score87.9 Script 783.0 Script 6
Range across scripts3.63.2

Spread is how much the Writing Score moves between a configuration's drafts, and lower is steadier. Green marks the better number in each row.

Scores by Script

Writing Score on each reference script. Gap is Claude Opus 5.5 (xhigh) minus GPT-5.6 Terra (ultra), and the higher score is in bold.

ScriptClaude Opus 5.5 (xhigh)GPT-5.6 Terra (ultra)Gap
Script 1 opinion / warning explainer84.481.1+3.3
Script 2 news-analysis / skeptical explainer85.682.1+3.5
Script 3 personal roadmap / opinion85.782.5+3.2
Script 4 short explainer86.279.8+6.4
Script 5 news-analysis / opinion explainer85.182.2+2.9
Script 6 founder announcement / personal origin story85.683.0+2.6
Script 7 personal technical walkthrough / agentic workflow case study87.980.6+7.3
Script 8 product release announcement / personal observation85.880.3+5.5
Script 9 career guide / hiring analysis87.981.8+6.1
Script 10 engineering process walkthrough / presentation adaptation84.380.6+3.7

Compare With Another Model

Pick a second model. A pair with its own written match-up opens that page, and any other pair opens in the comparison tool.

Nearby Match-Ups by Best Writing Elo

Every configuration wrote five drafts for each of the same 10 briefs, from the same prompt. Generation was closed-book, with the prompt as the only source and no web search or other tools. Each judge scores every draft once: Claude Opus 5.5 at medium effort through the Claude Code subscription, GPT-6.1 Sol at medium reasoning through the Codex CLI, GLM 5.3 Flash through OpenRouter, and Jev 1.13, a typed judge from TypeSafe that answers each metric as a multiple-choice question. The channel's finished script is their reference for the voice, and it gets no score on the board itself.

← All comparisons