Daily digest, 10 Sept 2026
1 new model, 7 new or updated benchmark results across 19 sources.
BenchLeader refreshed 1212 models from 19 of 19 sources. 1 new model, 7 new or updated benchmark results across 19 sources.
New models
- Deepseek v4 1 Flash (DeepSeek) appeared on 0 benchmarks. (details)
Score revisions
- Qwen3.8-Flash-Next on AA Intelligence Index: 42.2 → 39.9. (details)
- GPT-5.6 Luna (high) on AA Intelligence Index: 32.9 → 32.4. (details)
- GPT-5.6 Terra (medium) on AA Intelligence Index: 32.8 → 30.4. (details)
- Qwen 3.8 27B (low) on AA Intelligence Index: 29.2 → 26.5. (details)
- Qwen 3.8 27B (medium) on AA Intelligence Index: 30.7 → 27.8. (details)
- Qwen3.6 35B-A3B on AA Intelligence Index: 22.3 → 18.8. (details)
- Granite 4.2 8B on AA Intelligence Index: 12.4 → 11.8. (details)
Speed changes
- GPT-6 Astra (max) time to first token changed from 7.00 s to 4.47 s. (details)
- GPT-6 Astra time to first token changed from 7.00 s to 4.47 s. (details)
- GPT-6 Astra (high) time to first token changed from 7.00 s to 4.47 s. (details)
- GPT-6 Astra (xhigh) time to first token changed from 7.00 s to 4.47 s. (details)
- GPT-5.6 Sol (high) time to first answer changed from 31.42 s to 39.91 s. (details)
- GPT-5.6 Sol (high) time to first token changed from 5.46 s to 3.31 s. (details)
- Claude Fable 5.1 (medium) time to first answer changed from 11.60 s to 17.28 s. (details)
- GPT-5.6 Sol (xhigh) time to first token changed from 5.46 s to 3.31 s. (details)
- GPT-6 Astra (medium) time to first token changed from 7.00 s to 4.47 s. (details)
- GPT-5.5 time to first token changed from 3.50 s to 5.66 s. (details)
- GPT-5.5 (high) time to first answer changed from 14.11 s to 23.02 s. (details)
- GPT-5.5 (high) time to first token changed from 3.50 s to 5.66 s. (details)
- GPT-5.6 Sol (medium) time to first token changed from 5.46 s to 3.31 s. (details)
- GPT-6 Astra (low) time to first token changed from 7.00 s to 4.47 s. (details)
- GPT-5.6 Sol (max) time to first token changed from 5.46 s to 3.31 s. (details)
- …and 230 more.
Today's top five
- GPT-6 Astra — 70.6
- Claude Fable 5 — 70.4
- Claude Fable 5.1 — 70.3
- Claude Fable 5.1 — 70.2
- Claude Opus 5 — 70.0
This digest is generated automatically from the day's data changes. Every line links to the page where you can check the numbers and their source.