Lab Comparison
Last updated , from board data generated at 17:52 UTC.
All 7 ranked Zhipu (GLM) configurations on ToneBench, side by side: Writing Elo, Writing Score, the nine metrics, and cost and time per script. A configuration is one model at one thinking level.
A configuration 200 Elo below the top is expected to score about 24% in its script-by-script comparisons with it, a draw counting as half. The Lowest Cost Within 200 Elo of the Top card names the cheapest configuration that stays that close.
Ranks in these notes are configuration ranks: each setting's place among all 151 configurations. The Highest Writing Elo card above gives the model rank.
Picks and Notes From Louis
If you only care about the score, GLM-5.3 (max) is the one. The pick I'd make is GLM-5.3 Flash (max). It's 92 Elo behind, and their ranges don't overlap. The price decides it for me: GLM-5.3 (max) costs 10.8 times as much. Neither is fast: 5.2 minutes per script for the Flash and 6.1 for GLM-5.3 (max). At $0.022 a script, I'd take the wait.
Below those two, I'd skip the rest. GLM-5 (thinking on) and GLM-5.2 (high) both cost more than GLM-5.3 Flash (max) and land at #89 and #91. GLM-4.5 Air (thinking on) is the cheapest GLM and the fastest, and it's also last in the family at #141.
Zhipu's GLM line is open weights top to bottom, and its best model is a serious writer: GLM-5.3 (max) ranks #22, the highest open-weights row on the whole board. At the other end, GLM-4.5 Air (thinking on) lands at #141, 1703 Elo lower. The jump happens at GLM-5.3. The older GLM-5 rows already sit far below it, and the GLM-4 generation is lower still, so with GLM the version you pick matters more than the model size. One disclosure: GLM 5.3 Flash is also one of the four judges, and the other three score every draft too.
Every row wrote the same 10 scripts for the same four blind judges, so these Writing Elo ratings sit on the leaderboard's scale. The rank column is the configuration rank, each setting's place among all 151 configurations on the board. Each row is a different model. The ± after each Elo is half the width of its 95% range, and scores run from 0 to 100. Cost is one full script at the API list price of the route used. Each name opens its scorecard, and the head-to-head comparisons cover models from different labs.
With the metric scores shown, the table scrolls sideways.
| Configuration Rank | Configuration | Writing Elo | Writing Score | Cost / Script | Time / Script | Tone & Voice Match | Writing Craft & Clarity | Substance, Accuracy & Value | Hook Strength | YouTube Best Practices | Continuity & Emotion | Slop Score | Length Adherence | Visual Cue Quality |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| #22 | GLM-5.3 (max) · open weights | 2142 ±35 | 82.4 | $0.240 | 6.1 min | 79.7 | 78.7 | 83.2 | 83.4 | 81.7 | 82.1 | 83.4 | 94.3 | 79.7 |
| #29 | GLM-5.3 Flash (max) · open weights | 2050 ±40 | 81.8 | $0.022 | 5.2 min | 78.0 | 78.3 | 82.8 | 82.8 | 80.9 | 81.3 | 83.1 | 97.1 | 78.0 |
| #89 | GLM-5 (thinking on) · open weights | 1528 ±36 | 76.7 | $0.116 | 6.1 min | 73.3 | 72.5 | 79.7 | 78.2 | 75.1 | 74.9 | 79.5 | 88.5 | 75.7 |
| #91 | GLM-5.2 (high) · open weights | 1479 ±44 | 75.8 | $0.068 | 1.4 min | 72.8 | 72.5 | 79.9 | 79.3 | 72.4 | 74.4 | 79.2 | 82.5 | 73.2 |
| #100 | GLM-4.7 (thinking on) · open weights | 1378 ±69 | 74.0 | $0.041 | 6.2 min | 72.1 | 71.2 | 77.7 | 79.6 | 73.4 | 73.3 | 78.1 | 71.0 | 70.3 |
| #102 | GLM-4.6 (thinking on) · open weights | 1347 ±48 | 74.3 | $0.045 | 7.0 min | 71.7 | 71.9 | 77.5 | 80.5 | 73.3 | 74.1 | 78.7 | 71.8 | 70.8 |
| #141 | GLM-4.5 Air (thinking on) · open weights | 439 ±40 | 59.5 | $0.0066 | 0.8 min | 52.7 | 62.3 | 62.7 | 72.6 | 58.5 | 60.2 | 66.4 | 45.2 | 57.0 |
Each comparison lists every thinking level of both models, with Writing Elo, cost and time side by side. The line under each pair gives the Writing Elo gap between the two best configurations. A pair outside the written match-ups opens in the comparison tool.