Particle-1 added, GPT-6.1 Sol up to #8 (14 Nov 2026)
Particle-1 joins the board; GPT-6.1 Sol moved #9 → #8. 1 new model, 2 top-10 rank changes, 16 price changes, 599 new benchmark results. Sources linked.
BenchLeader refreshed 1705 models from 29 of 29 data feeds (benchmark publishers, aggregators and price lists). 1 new model, 2 top-10 rank changes, 16 price changes, 599 new or updated benchmark results across 29 data feeds.
New models
- Particle-1 (Particle AI) is now listed at $0.17/M blended; no independent results yet. (details)
Movement in the top 10
- GPT-6.1 Sol (max) moved from #9 to #8 in the BenchLeader Index. (details)
- Claude Fable 5.1 (max) moved from #8 to #9 in the BenchLeader Index. (details)
Price changes
- Qwen3.8 27B input price moved from $0.45/M to $0.42/M (7% down). (details)
- Nemotron 3 Ultra 550B A55B (thinking) output price moved from $2.70/M to $2.40/M (11% down). (details)
- Muse Glimmer (high) input price moved from $0.30/M to $0.32/M (7% up). (details)
- Muse Glimmer (high) output price moved from $1.20/M to $1.35/M (13% up). (details)
- Muse Glimmer (high) cost per task moved from $0.17/M to $0.18/M (5% up). (details)
- Nemotron 3.5 Super VL input price moved from $0.18/M to $0.20/M (11% up). (details)
- Nemotron 3.5 Super VL cost per task moved from $0.52/M to $0.56/M (7% up). (details)
- DeepSeek V4 Pro 0424 (max) input price moved from $1.31/M to $1.50/M (15% up). (details)
- MiMo-V2.5-Pro (thinking) input price moved from $0.39/M to $0.35/M (10% down). (details)
- MiMo-V2.5-Pro (thinking) output price moved from $0.78/M to $0.70/M (10% down). (details)
- Nemotron 3.5 Super VL (no reasoning) input price moved from $0.18/M to $0.20/M (11% up). (details)
- Nemotron 3.5 Super VL (no reasoning) cost per task moved from $0.52/M to $0.56/M (7% up). (details)
- Nemotron 3.5 Super VL (medium) input price moved from $0.18/M to $0.20/M (11% up). (details)
- Nemotron 3.5 Super VL (medium) cost per task moved from $0.52/M to $0.56/M (7% up). (details)
- Nemotron 3.5 Super VL (high) input price moved from $0.18/M to $0.20/M (11% up). (details)
- Nemotron 3.5 Super VL (high) cost per task moved from $0.52/M to $0.56/M (7% up). (details)
New benchmark results
- AIME 2025 (AA): 177 new results, including Gemini 3 Pro, GPT-5.2, Claude Opus 4.5… (board)
- GDPval-AA v2.1: 142 new results, including Claude Fable 5.1, GPT-5.6 Sol, Claude Fable 5… (board)
- Global-MMLU-Lite (AA): 83 new results, including Gemini 3 Flash, GPT-5, Claude Opus 4.6… (board)
- AA-Briefcase v1.1: 80 new results, including Claude Fable 5.1, Muse Spark, Claude Opus 5… (board)
- DeepSec-Bench (AA): 4 new results, including GPT-5.6 Terra, Muse Spark 1.3, Mistral Large 4… (board)
- Terminal-Bench: 3 new results, including Gemini 4 Carbon, Gemini 4 Argon, Muse Wonder (board)
- AA Cyber Index: 3 new results, including Muse Spark 1.3, Mistral Large 4, GPT-6 Sol (Daybreak Blue, max) (board)
- CWE-Bench (AA): 3 new results, including Muse Spark 1.3, Mistral Large 4, GPT-6 Sol (Daybreak Blue, max) (board)
- CyberGym end-to-end (AA): 3 new results, including Muse Spark 1.3, Mistral Large 4, GPT-6 Sol (Daybreak Blue, max) (board)
- Design Arena Agentic: 2 new results, including Hy3, Gemini 4 Hydrogen (board)
- AA Intelligence Index v5.0.1: 1 new result, including Nemotron 3 Ultra 550B A55B (board)
Score revisions
- Claude Fable 5.2 (max) on Coding-agent share: 0.6% → 1.3%. (details)
- Claude Fable 5.2 (xhigh) on Coding-agent share: 0.6% → 1.3%. (details)
- Claude Fable 5.2 (high) on Coding-agent share: 0.6% → 1.3%. (details)
- GPT-6 Astra (max) on Design Arena Agentic: 1393 → 1401. (details)
- GPT-6.1 Sol (max) on Design Arena Agentic: 1420 → 1370. (details)
- Claude Fable 5.1 on Design Arena Agentic: 1327 → 1333. (details)
- Claude Fable 5 on Design Arena Agentic: 1266 → 1271. (details)
- Claude Fable 5.2 (medium) on Coding-agent share: 0.6% → 1.3%. (details)
- Claude Fable 5.2 on Coding-agent share: 0.6% → 1.3%. (details)
- Claude Opus 5.5 (high) on Design Arena Agentic: 1408 → 1449. (details)
- Claude Fable 5.2 (low) on Coding-agent share: 0.6% → 1.3%. (details)
- Gemini 4 Argon on Design Arena Agentic: 1262 → 1271. (details)
- GPT-6.1 Sol (xhigh) on GDPval-AA v2.1: 1555 → 1563. (details)
- Gemini 4 Carbon on Design Arena Website: 1324 → 1319. (details)
- GPT-6.1 Sol on Design Arena Agentic: 1252 → 1259. (details)
- Gemini 3.5 Flash (high) on GameBench 2 by reasoning setting: 62.2 → 63.2. (details)
- Claude Sonnet 5.5 (xhigh) on GDPval-AA v2.1: 1735 → 1743. (details)
- Kimi K3 (max) on GameBench 2 by reasoning setting: 44.8 → 45.7. (details)
- Muse Spark 1.3 (max) on Design Arena Agentic: 1274 → 1279. (details)
- Muse Wonder (max) on Design Arena Agentic: 1159 → 1168. (details)
- …and 78 more.
Speed changes
- Claude Fable 5.1 time to first answer changed from 47.94 s to 82.17 s. (details)
- GPT-5.5 time to first token changed from 1.59 s to 3.08 s. (details)
- GPT-5.4 time to first token changed from 2.83 s to 1.69 s. (details)
- Muse Wonder output speed changed from 39 tok/s to 70 tok/s. (details)
- Muse Wonder time to first token changed from 7.15 s to 3.87 s. (details)
- Gemini 4 Carbon output speed changed from 30 tok/s to 43 tok/s. (details)
- Gemini 4 Argon output speed changed from 34 tok/s to 56 tok/s. (details)
- Muse Spark 1.2 time to first token changed from 2.32 s to 3.31 s. (details)
- Kimi K3.1 Pro time to first token changed from 1.08 s to 2.00 s. (details)
- Inkling 2 Small time to first token changed from 0.22 s to 0.33 s. (details)
- DeepSeek V4 Flash time to first token changed from 0.46 s to 0.87 s. (details)
- MiniMax M3 time to first token changed from 0.76 s to 1.29 s. (details)
- DeepSeek V4 Pro output speed changed from 53 tok/s to 76 tok/s. (details)
- DeepSeek V4 Pro time to first token changed from 0.97 s to 1.55 s. (details)
- GLM 5.1 time to first token changed from 2.23 s to 1.28 s. (details)
- …and 53 more.
Today's top five
- Claude Fable 5.2 (max) — 72.7
- Gemini 4 Carbon (high) — 68.9
- Claude Opus 5.5 (max) — 68.4
- GPT-6 Astra (max) — 67.3
- Claude Fable 5.1 (max) — 66.7
More: New AI model releases this week, and the full LLM leaderboard 2026.
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.