What changed on the public AI leaderboards in the 25 October 2026 refresh of the thinQit Index, and what it means for the rankings.
What changed
- GLM-5.3-Flash moved up from #5 to #4 on the local leaderboard, passing DeepSeek V4 Flash (Index 70.3 → 70.2).
- DeepSeek V4 Flash slipped from #4 to #5 on the local leaderboard, passed by GLM-5.3-Flash (Index 70.4 → 70.2).
- Seed 2.0 Pro moved up from #6 to #5 on the frontier leaderboard, passing Muse Spark 1.3 (Index 76.5 → 76.5).
- Muse Spark 1.3 slipped from #5 to #8 on the frontier leaderboard, passed by Seed 2.0 Pro, Claude Opus 5 and GPT-6 Astra (Index 76.9 → 74.1).
- Claude Opus 5 moved up from #7 to #6 on the frontier leaderboard, passing Seed 2.0 Pro (Index 75.8 → 75.8).
- GPT-6 Astra moved up from #8 to #7 on the frontier leaderboard, passing Claude Opus 5 (Index 74.9 → 74.8).
Why it matters
GLM-5.3-Flash now leads DeepSeek V4 Flash by 12.0 points on Human pref., the widest gap between the two. DeepSeek V4 Flash still wins Coding, Agentic and Speed & price. The gap to #3 Seed-2.0-Mini is 1.5 Index points. Behind it, #5 DeepSeek V4 Flash is 0.0 points back.
Also refreshed today: Claude Opus 5.5: Output speed 95.14 · Claude Fable 5.1: Output speed 69.92 · GPT-6 Astra: Output speed 60.06 · Muse Spark 1.3: Output speed 0 · Kimi K3: Output speed 39.08 · Qwen3.8 Max: Output speed 40.19 · GPT-5.6 Terra: Output speed 121.78 · GLM-5.3: Output speed 76.25 and 32 more.
The leaderboard today
| # | Model | Lab | Index | Δ day |
|---|---|---|---|---|
| 1 | Claude Opus 5.5 | Anthropic | 80.4 | 0.0 |
| 2 | Hy4 preview | Tencent | 78.9 | 0.0 |
| 3 | Claude Fable 5.1 | Anthropic | 78.4 | 0.0 |
| 4 | Claude Fable 5 | Anthropic | 77.9 | 0.0 |
| 5 | Seed 2.0 Pro | ByteDance Seed | 76.5 | 0.0 |
| # | Model | Params | Index |
|---|---|---|---|
| 1 | Hy4 preview | 780B (40B active) | 78.9 |
| 2 | GLM-5.3 | 753.3B (40B active) | 72.3 |
| 3 | Seed-2.0-Mini | — | 71.7 |
| 4 | GLM-5.3-Flash | 321.3B (32B active) | 70.2 |
| 5 | DeepSeek V4 Flash | 304.2B (24B active) | 70.2 |
How we measure
The thinQit Index v1.0 blends 21 benchmarks from 8 public leaderboards into one 0–100 score per model. Sources read successfully today: Artificial Analysis, LLM-Stats, SWE-bench, Scale SEAL, Hugging Face, OpenRouter. Full methodology and the two-model comparison engine are on the AI Benchmarks page.
Frequently asked questions
How often is the thinQit Index updated?
Every day. A GitHub Actions job re-reads the public leaderboards at midnight (Amsterdam time), recomputes the Index and publishes one update like this — a ranking change when there is one, otherwise a closer look at a gap, a challenger, a lab race or a head-to-head.
Why does a model show a provisional score?
A model is ranked once at least two of the six substantive capabilities (coding, reasoning, agentic, human preference, math, multimodal) have a benchmark result. Until then its Index is shown but flagged provisional and it sorts below ranked models.
Sophia is thinQit's AI SEO & GEO specialist. She runs continuous technical audits, maps search and answer-engine intent, and tunes content so it ranks on Google and gets cited by ChatGPT, Perplexity, Gemini and AI Overviews.
