The Latent Score: 2026-09-05
Release tli-2026-v4.2-2026-09-05. Methodology tli-2026-v4.2.
As of 2026-09-05, Fable 5.1 leads The Latent Score on the point estimate at 128.1 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 | 128.1 | 122.7–134.2 | 53 |
| 2 | GPT-6 Astra | 124.0 | 113.2–149.6 | 40 |
| 3 | Opus 5 | 121.4 | 114.9–126.2 | 81 |
| 4 | Fable 5 | 118.0 | 112.9–122.6 | 93 |
| 5 | Muse Spark 1.3 | 112.1 | 105.7–119.1 | 19 |
| 6 | GPT-5.6 Sol | 108.1 | 103.6–114.4 | 99 |
| 7 | Kimi K3 | 104.7 | 98.5–109.3 | 84 |
| 8 | GLM-5.3 | 103.3 | 95.5–108.6 | 52 |
| 9 | Muse Spark 1.2 | 102.3 | 93.3–107.4 | 46 |
| 10 | Gemini 3.8 Flash | 102.3 | 93.7–109.9 | 46 |
| 11 | Grok 4.6 | 101.9 | 96.9–108.5 | 54 |
| 12 | Qwen3.8-Max | 96.9 | 89.5–101.5 | 67 |
| 13 | Gemini 3.7 Flash | 94.9 | 86–104.7 | 56 |
| 14 | GPT-5.6 Terra | 94.5 | 90.3–102.6 | 70 |
| 15 | DeepSeek V4 Pro | 89.4 | 82.6–97.5 | 53 |
| 16 | Sonnet 5 | 88.6 | 82.1–93.9 | 64 |
| 17 | GPT-5.6 Luna | 86.0 | 81.7–94.6 | 67 |
| 18 | DeepSeek V4 Flash | 72.9 | 67.7–83 | 51 |
| 19 | Qwen3.8-27B | 69.8 | 59.6–80.1 | 49 |
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. 100 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-v4.2. 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-v4.2-2026-09-05, dataset dated 2026-09-05. Permanent release page.
Download public scores with definitions (JSON) · Data usage terms