The Latent Score: 2026-09-03
Release tli-2026-v4.1-2026-09-03b. Methodology tli-2026-v4.1.
As of 2026-09-03, Fable 5.1 leads The Latent Score on the point estimate at 130.5 points among 19 ranked models. Rankings describe published capability evidence, not a statistically proven winner or the best model for every workload.
This page preserves a specific publication. Its data does not change when a new release is published.
Published rankings
| Rank | Model and evidence | Score (points) | Stability range (points) | Contributing results |
|---|---|---|---|---|
| 1 | Fable 5.1 | 130.5 | 121.9–137.5 | 28 |
| 2 | GPT-6 Astra | 122.4 | 108.9–151.8 | 16 |
| 3 | Opus 5 | 121.9 | 115.7–128 | 66 |
| 4 | Fable 5 | 118.1 | 113–122.7 | 79 |
| 5 | Muse Spark 1.3 | 109.5 | 101.2–117.9 | 11 |
| 6 | GPT-5.6 Sol | 107.9 | 103.1–113.9 | 78 |
| 7 | Kimi K3 | 105.5 | 98.8–110.2 | 69 |
| 8 | GLM-5.3 | 105.1 | 94–111.3 | 30 |
| 9 | Muse Spark 1.2 | 105.0 | 93.9–110.1 | 37 |
| 10 | Gemini 3.8 Flash | 101.4 | 92.2–113.3 | 33 |
| 11 | Grok 4.6 | 101.4 | 96.2–108.1 | 44 |
| 12 | Qwen3.8-Max | 98.2 | 90.6–102.6 | 57 |
| 13 | Gemini 3.7 Flash | 93.9 | 84–105.3 | 28 |
| 14 | GPT-5.6 Terra | 93.7 | 89.2–101.6 | 54 |
| 15 | DeepSeek V4 Pro | 89.4 | 83.1–98.4 | 39 |
| 16 | Sonnet 5 | 88.2 | 81.1–93.6 | 53 |
| 17 | GPT-5.6 Luna | 86.5 | 82.1–94 | 53 |
| 18 | DeepSeek V4 Flash | 72.6 | 67–81.9 | 47 |
| 19 | Qwen3.8-27B | 71.5 | 59.4–82.6 | 26 |
Methodology and limitations in this release
The scale is anchored to mean 100, SD 15 across 18 frozen reference models. It is not human IQ or a percentage. 80 benchmarks contribute across 7 domains.
This older release did not preserve a written methodology in its public snapshot. Its recorded scores and scale are archived here; current methodology documentation must not be assumed to describe this historical calculation.
Reading this release
Compare model scores alongside their stability ranges and domain coverage. A higher point estimate does not establish superiority for every task. The linked model records disclose contributing and excluded results and their published configurations.
This release uses methodology version tli-2026-v4.1. Comparisons with another release should check both its methodology version and evidence date; a new data release does not necessarily mean the scoring method changed.
Citation and downloads
The Latent. “The Latent Score.” Release tli-2026-v4.1-2026-09-03b, dataset dated 2026-09-03. Permanent release page.
Download public scores with definitions (JSON) · Data usage terms