| GLM-5.3 | Claude Fable 5.2 | |
|---|---|---|
| CTO Index | 62.0 (#21) | 95.5 (#1) |
| Provider | Z.ai | Anthropic |
| Weights | Open weights | Closed |
| Price $/M in / out | $1.40 / $4.40 | $8 / $40 |
| Context window | 1M | 1M |
| License | MIT | Proprietary |
| Terminal-Bench 4.1 | 41.2% | 65.8% |
| WebDev Arena | 1623 | 1824 |
| ARC-AGI-2 | — | 93.8% |
| Text Arena | 1482 | 1503 |
| ARC-AGI-3 | — | 81.2% |
Snapshot 22 November 2026 · Scores copied from each benchmark's own leaderboard · How the CTO Index works
Frequently asked questions
Is GLM-5.3 better than Claude Fable 5.2?
Claude Fable 5.2 scores higher on the CTO Index (95.5 vs 62.0). Across the 3 benchmarks both have results, GLM-5.3 wins 0 and Claude Fable 5.2 wins 3 (Terminal-Bench, WebDev, Text Arena). GLM-5.3 costs 87% less per blended token.
Which is better for coding agents, GLM-5.3 or Claude Fable 5.2?
On Terminal-Bench 4.1, Claude Fable 5.2 scores 65.8% against GLM-5.3's 41.2%. Differences under about three points are within the error bars.
How do GLM-5.3 and Claude Fable 5.2 compare on price?
GLM-5.3 lists at $1.40 / $4.40 and Claude Fable 5.2 at $8 / $40 per million input / output tokens.