Release watch
New AI model releases: the last 30 days
What arrived, what got measured, and who is due. Rebuilt once a day from the day’s data.
As of 25 Oct 2026, 9 new models arrived this week: Kimi K3.1 Pro (Moonshot AI, 23 Oct), Nemotron 3.5 Bolt E4B 4GB (NVIDIA, 23 Oct), Nemotron Super VL 3.5 NVFP4 (NVIDIA, 23 Oct), Smaug Nemotron 3.5 NVFP4 (NVIDIA, 23 Oct) and more.
Data as of 25 Oct 2026
New this week
| Model | Released | Index | Rank | Price $/M | Status |
|---|---|---|---|---|---|
| 23 Oct 2026 | 66.9 | #28 | $4.00 | Ranked | |
| 23 Oct 2026 | – | – | – | First results arriving | |
| seen 23 Oct 2026 | – | – | – | First results arriving | |
| seen 23 Oct 2026 | – | – | – | First results arriving | |
| Spark X2.6Synlan | seen 23 Oct 2026 | – | – | – | First results arriving |
| 23 Oct 2026 | – | – | $8.00 | Released; awaiting independent results | |
| seen 21 Oct 2026 | – | – | – | Listed; awaiting independent results | |
| 19 Oct 2026 | 61.9 | #92 | $0.262 | Ranked | |
| 19 Oct 2026 | – | – | $0.163 | Released; awaiting independent results |
Released in the last 30 days
Every model released, or first listed with no release date, in the 30 days to 25 Oct 2026, newest first.
| Model | Provider | Date | Index | Rank | Results |
|---|---|---|---|---|---|
| Liquid AI | listed 25 Oct 2026 | – | – | 0 | |
| Moonshot AI | 23 Oct 2026 | 66.9 | #28 | 9 | |
| NVIDIA | 23 Oct 2026 | – | – | 1 | |
| NVIDIA | listed 23 Oct 2026 | – | – | 1 | |
| NVIDIA | listed 23 Oct 2026 | – | – | 1 | |
| Spark X2.6 | Synlan | listed 23 Oct 2026 | – | – | 1 |
| Moonshot AI | 23 Oct 2026 | – | – | 0 | |
| OpenAI | listed 21 Oct 2026 | – | – | 0 | |
| Cohere | listed 20 Oct 2026 | – | – | 2 | |
| MiniMax | 19 Oct 2026 | 61.9 | #92 | 7 | |
| Inception | 19 Oct 2026 | – | – | 0 | |
| DeepSeek | 17 Oct 2026 | 64.0 | #58 | 16 | |
| ByteDance | listed 17 Oct 2026 | – | – | 1 | |
| Salesforce | listed 17 Oct 2026 | – | – | 0 | |
| Zhipu AI | listed 16 Oct 2026 | – | – | 1 | |
| Meta | 15 Oct 2026 | 62.0 | #90 | 7 | |
| Meta | 15 Oct 2026 | 58.6 | #150 | 6 | |
| NVIDIA | 15 Oct 2026 | 50.4 | #373 | 7 | |
| NVIDIA | 15 Oct 2026 | – | – | 4 | |
| NVIDIA | 15 Oct 2026 | – | – | 4 | |
| Liquid AI | listed 14 Oct 2026 | – | – | 0 | |
| Meta | 12 Oct 2026 | 70.6 | #3 | 7 | |
| Meta | 12 Oct 2026 | 67.5 | #22 | 15 | |
| Anthropic | 7 Oct 2026 | 61.5 | #95 | 8 | |
| Mistral AI | 6 Oct 2026 | 56.7 | #196 | 18 | |
| 6 Oct 2026 | – | – | 0 | ||
| Reflection | 5 Oct 2026 | 54.9 | #241 | 6 | |
| Reflection | 5 Oct 2026 | 51.2 | #350 | 9 | |
| Ant Group | 1 Oct 2026 | – | – | 4 | |
| 30 Sept 2026 | 70.7 | #2 | 21 | ||
| OpenAI | 29 Sept 2026 | 69.9 | #7 | 24 | |
| Anthropic | listed 29 Sept 2026 | – | – | 0 | |
| Anthropic | listed 29 Sept 2026 | – | – | 0 | |
| Inception | 29 Sept 2026 | – | – | 0 | |
| Resonant 1 | Unknown | listed 29 Sept 2026 | – | – | 0 |
| Anthropic | 28 Sept 2026 | 69.4 | #10 | 15 | |
| SpaceXAI | listed 27 Sept 2026 | 53.7 | #277 | 8 |
Who is due
Expected windows from each provider’s release cadence over the last two years. An estimate from dates only, not from announcements.
| Provider | Last release | Cadence | Expected window |
|---|---|---|---|
| 30 Apr 2025 | every 62 days | 6 Jun 2025 to 25 Jul 2025overdue | |
| 12 Dec 2025 | every 112 days | 17 Feb 2026 to 17 May 2026overdue | |
| 2 Dec 2025 | every 148 days | 28 Feb 2026 to 27 Jun 2026overdue | |
| 8 Jan 2026 | every 92 days | 4 Mar 2026 to 16 May 2026overdue | |
| 14 Feb 2026 | every 83 days | 4 Apr 2026 to 10 Jun 2026overdue | |
| 30 Jan 2026 | every 156 days | 3 May 2026 to 5 Sept 2026overdue | |
| China Mobile | 9 Jul 2026 | every 56 days | 11 Aug 2026 to 25 Sept 2026overdue |
| 21 Jul 2026 | every 65 days | 29 Aug 2026 to 20 Oct 2026overdue | |
| 4 Aug 2026 | every 68 days | 13 Sept 2026 to 7 Nov 2026 | |
| 17 Sept 2026 | every 7 days | 21 Sept 2026 to 26 Sept 2026overdue | |
| 18 Sept 2026 | every 28 days | 4 Oct 2026 to 27 Oct 2026 | |
| 21 Sept 2026 | every 26 days | 6 Oct 2026 to 27 Oct 2026 | |
| 29 Sept 2026 | every 22 days | 12 Oct 2026 to 29 Oct 2026 | |
| 1 Oct 2026 | every 19 days | 12 Oct 2026 to 27 Oct 2026 | |
| 6 Oct 2026 | every 15 days | 15 Oct 2026 to 27 Oct 2026 | |
| 21 Sept 2026 | every 41 days | 15 Oct 2026 to 17 Nov 2026 | |
| 22 Sept 2026 | every 41 days | 16 Oct 2026 to 18 Nov 2026 | |
| 6 Oct 2026 | every 28 days | 22 Oct 2026 to 14 Nov 2026 | |
| 18 Sept 2026 | every 59 days | 23 Oct 2026 to 9 Dec 2026 | |
| 9 Sept 2026 | every 76 days | 24 Oct 2026 to 24 Dec 2026 | |
| 12 Aug 2026 | every 125 days | 26 Oct 2026 to 3 Feb 2027 | |
| 7 Oct 2026 | every 39 days | 30 Oct 2026 to 30 Nov 2026 | |
| 25 Aug 2026 | every 118 days | 3 Nov 2026 to 6 Feb 2027 | |
| 15 Oct 2026 | every 40 days | 8 Nov 2026 to 10 Dec 2026 | |
| 17 Oct 2026 | every 37 days | 8 Nov 2026 to 7 Dec 2026 | |
| 7 Sept 2026 | every 105 days | 9 Nov 2026 to 1 Feb 2027 | |
| 19 Oct 2026 | every 38 days | 10 Nov 2026 to 11 Dec 2026 | |
| 23 Oct 2026 | every 36 days | 13 Nov 2026 to 12 Dec 2026 | |
| 19 Oct 2026 | every 57 days | 22 Nov 2026 to 6 Jan 2027 | |
| 23 Oct 2026 | every 56 days | 25 Nov 2026 to 9 Jan 2027 |
First results this week
Benchmarks that published a first score for a model in the last seven days.
- GPT-6 Sol
- Harvey's Legal Agent Benchmark v1: 1.7%23 Oct 2026
- Vals RSI Index: 28.1%23 Oct 2026
- Vals Science Index: 45%23 Oct 2026
- MLCR: 32.2%21 Oct 2026
- MLCR: 29.4%21 Oct 2026
- MLCR: 28.9%21 Oct 2026
- and 83 more
- DeepSeek V4 Flash
- Harvey's Legal Agent Benchmark v1: 8.3%23 Oct 2026
- LiveBench 2.0: 57.6%23 Oct 2026
- LiveBench Long Agentic: 36.6%23 Oct 2026
- LiveBench 2.0 Reasoning: 70.6%23 Oct 2026
- LiveBench 2.0 Coding: 77%23 Oct 2026
- LiveBench Browser: 0%23 Oct 2026
- and 51 more
- Claude Sonnet 5.5
- ClaimBench: 100%23 Oct 2026
- Code Migration v1.1: 86.9%23 Oct 2026
- Excel Modeling Benchmark v1.1: 75.9%23 Oct 2026
- Finance Agent v2.1: 67.3%23 Oct 2026
- Vals Finance Index: 64.5%23 Oct 2026
- Harvey's Legal Agent Benchmark v1: 2.9%23 Oct 2026
- and 46 more
- Grok 4.6
- Harvey's Legal Agent Benchmark v1: 15.8%23 Oct 2026
- Vals Science Index: 38.1%23 Oct 2026
- Vals RSI Index: 25.5%23 Oct 2026
- MLCR: 18.3%21 Oct 2026
- MLCR: 17.2%21 Oct 2026
- MLCR: 15%21 Oct 2026
- and 45 more
- DeepSeek V4 Pro
- Code Migration v1.1: 52%23 Oct 2026
- Excel Modeling Benchmark v1.1: 65.1%23 Oct 2026
- Finance Agent v2.1: 57.1%23 Oct 2026
- Vals Finance Index: 52.9%23 Oct 2026
- Harvey's Legal Agent Benchmark v1: 7.5%23 Oct 2026
- Vals Router Index: 53%23 Oct 2026
- and 44 more
- Gemini 3.7 Flash
- Harvey's Legal Agent Benchmark v1: 8.8%23 Oct 2026
- Vals RSI Index: 18.3%23 Oct 2026
- ForecastBench: 66.2%20 Oct 2026
- LiveBench 2.0: 67.2%19 Oct 2026
- LiveBench Long Agentic: 28.3%19 Oct 2026
- LiveBench 2.0 Reasoning: 87.8%19 Oct 2026
- and 39 more
- Claude Haiku 5.5
- Arena Text: 142523 Oct 2026
- Arena Hard Prompts: 145223 Oct 2026
- Arena Coding: 148523 Oct 2026
- Arena Maths: 142923 Oct 2026
- Arena Creative Writing: 136123 Oct 2026
- Arena Instruction Following: 142923 Oct 2026
- and 38 more
- GPT-6 Luna
- ClaimBench: 99.2%23 Oct 2026
- Code Migration v1.1: 46.7%23 Oct 2026
- Excel Modeling Benchmark v1.1: 70%23 Oct 2026
- Finance Agent v2.1: 56.8%23 Oct 2026
- Vals Finance Index: 53.1%23 Oct 2026
- Harvey's Legal Agent Benchmark v1: 2.9%23 Oct 2026
- and 37 more
- Claude Opus 5
- Vals RSI Index: 33%23 Oct 2026
- Harvey's Legal Agent Benchmark v1: 6.7%23 Oct 2026
- Vals Science Index: 47.9%23 Oct 2026
- ForecastBench: 66.5%20 Oct 2026
- LiveBench 2.0: 82.1%19 Oct 2026
- LiveBench Long Agentic: 62.1%19 Oct 2026
- and 36 more
- GPT-6.1 Sol
- ClaimBench: 100%23 Oct 2026
- Code Migration v1.1: 77.3%23 Oct 2026
- Excel Modeling Benchmark v1.1: 71.1%23 Oct 2026
- Finance Agent v2.1: 61.8%23 Oct 2026
- Vals Finance Index: 55.8%23 Oct 2026
- Harvey's Legal Agent Benchmark v1: 5.4%23 Oct 2026
- and 32 more
- Claude Opus 5.5
- EQ-Bench 4: 133925 Oct 2026
- Robo-bench: 14.6%23 Oct 2026
- Vals RSI Index: 37.3%23 Oct 2026
- ClaimBench: 100%23 Oct 2026
- Code Migration v1.1: 81%23 Oct 2026
- Excel Modeling Benchmark v1.1: 75.8%23 Oct 2026
- and 31 more
- GPT-5.6 Sol
- AA-Briefcase v1.1: 147925 Oct 2026
- Harvey's Legal Agent Benchmark v1: 2.5%23 Oct 2026
- Vals RSI Index: 23.9%23 Oct 2026
- Vals Science Index: 41.5%23 Oct 2026
- ForecastBench: 65.5%20 Oct 2026
- ForecastBench: 66.5%20 Oct 2026
- and 28 more
- Gemini 4 Argon
- EQ-Bench 4: 124025 Oct 2026
- Code Migration v1.1: 70.8%23 Oct 2026
- Excel Modeling Benchmark v1.1: 74.8%23 Oct 2026
- Finance Agent v2.1: 67.7%23 Oct 2026
- Vals Finance Index: 64.9%23 Oct 2026
- Harvey's Legal Agent Benchmark v1: 19.6%23 Oct 2026
- and 26 more
- GPT-5.6 Luna
- AA-Briefcase v1.1: 133925 Oct 2026
- Harvey's Legal Agent Benchmark v1: 1.3%23 Oct 2026
- Vals Science Index: 25.3%23 Oct 2026
- MLCR: 8.9%21 Oct 2026
- LiveBench 2.0: 66.1%19 Oct 2026
- LiveBench Long Agentic: 30.4%19 Oct 2026
- and 25 more
- DeepSeek V4.1 Flash 1017
- MLCR: 26.7%23 Oct 2026
- AA Cyber Index: 56.823 Oct 2026
- CWE-Bench (AA): 69.2%23 Oct 2026
- DeepSec-Bench (AA): 24.9%23 Oct 2026
- CyberGym end-to-end (AA): 76.3%23 Oct 2026
- Code Migration v1.1: 67.5%23 Oct 2026
- and 25 more
- GPT-5.6 Terra
- Harvey's Legal Agent Benchmark v1: 0.8%23 Oct 2026
- MLCR: 14.4%21 Oct 2026
- LiveBench 2.0: 75.0%19 Oct 2026
- LiveBench Long Agentic: 48.4%19 Oct 2026
- LiveBench 2.0 Reasoning: 90.6%19 Oct 2026
- LiveBench 2.0 Coding: 86.2%19 Oct 2026
- and 23 more
- Gemini 3.5 Flash
- Harvey's Legal Agent Benchmark v1: 2.5%23 Oct 2026
- AA Intelligence Index v4.3.2: 33.618 Oct 2026
- GPQA Diamond (AA): 92.1%18 Oct 2026
- Humanity's Last Exam (AA): 41.3%18 Oct 2026
- IFBench: 74.6%18 Oct 2026
- AA-LCR: 74.3%18 Oct 2026
- and 22 more
- Grok 4.7
- Code Migration v1.1: 53.2%23 Oct 2026
- Excel Modeling Benchmark v1.1: 71.5%23 Oct 2026
- Finance Agent v2.1: 60.6%23 Oct 2026
- Vals Finance Index: 58.5%23 Oct 2026
- Harvey's Legal Agent Benchmark v1: 12.5%23 Oct 2026
- Robo-bench: 0.2%23 Oct 2026
- and 22 more
- GPT-5.5
- AA-Briefcase v1.1: 113625 Oct 2026
- Harvey's Legal Agent Benchmark v1: 3.8%23 Oct 2026
- Vals RSI Index: 16.5%23 Oct 2026
- MLCR: 25%21 Oct 2026
- MLCR: 22.2%21 Oct 2026
- ForecastBench: 67%20 Oct 2026
- and 21 more
- Claude Sonnet 5
- AA-Briefcase v1.1: 135725 Oct 2026
- Harvey's Legal Agent Benchmark v1: 5%23 Oct 2026
- MLCR: 4.4%21 Oct 2026
- MLCR: 5.6%21 Oct 2026
- MLCR: 5%21 Oct 2026
- MLCR: 6.1%21 Oct 2026
- and 21 more
- Gemini 3.8 Flash
- Code Migration v1.1: 38.7%23 Oct 2026
- Excel Modeling Benchmark v1.1: 72.9%23 Oct 2026
- Finance Agent v2.1: 65.8%23 Oct 2026
- Vals Finance Index: 60.5%23 Oct 2026
- Harvey's Legal Agent Benchmark v1: 10%23 Oct 2026
- Vals RSI Index: 19.9%23 Oct 2026
- and 21 more
- Beam 1.0
- MLCR: 17.8%23 Oct 2026
- Code Migration v1.1: 22.9%23 Oct 2026
- Excel Modeling Benchmark v1.1: 64.7%23 Oct 2026
- Finance Agent v2.1: 51.8%23 Oct 2026
- Vals Finance Index: 46.5%23 Oct 2026
- Vals Science Index: 22%23 Oct 2026
- and 20 more
- Muse Wonder
- EQ-Bench 4: 132425 Oct 2026
- MLCR: 22.2%23 Oct 2026
- Code Migration v1.1: 72.4%23 Oct 2026
- Excel Modeling Benchmark v1.1: 72.5%23 Oct 2026
- Finance Agent v2.1: 65.6%23 Oct 2026
- Vals Finance Index: 61.2%23 Oct 2026
- and 19 more
- GPT-6 Astra
- Code Migration v1.1: 80.6%23 Oct 2026
- Excel Modeling Benchmark v1.1: 72.7%23 Oct 2026
- Finance Agent v2.1: 60.4%23 Oct 2026
- Vals Finance Index: 55.6%23 Oct 2026
- Harvey's Legal Agent Benchmark v1: 5.4%23 Oct 2026
- Robo-bench: 23.3%23 Oct 2026
- and 19 more
Questions people ask
- What AI models were released this week?
- In the seven days to 25 Oct 2026: Kimi K3.1 Pro (Moonshot AI, 23 Oct 2026), Nemotron 3.5 Bolt E4B 4GB (NVIDIA, 23 Oct 2026), Nemotron Super VL 3.5 NVFP4 (NVIDIA, 23 Oct 2026), Smaug Nemotron 3.5 NVFP4 (NVIDIA, 23 Oct 2026), Spark X2.6 (Synlan, 23 Oct 2026), Kimi K3.1 Pro Highspeed (Moonshot AI, 23 Oct 2026), GPT-5.6 Sol Keenable (OpenAI, 21 Oct 2026) and MiniMax M3.1 Flash (MiniMax, 19 Oct 2026), and 1 more.
- What new AI models came out today?
- No new releases in today's data run (25 Oct 2026); the latest was Lfm 2.5 2 6B on 25 Oct 2026.
- What is the best new AI model this month?
- Of the models released in the 30 days to 25 Oct 2026, Gemini 4 Argon (high effort) by Google ranks highest, at 70.7 on the BenchLeader Index (top 10 overall).
- Why does a new model have no score yet?
- The BenchLeader Index needs at least five independent results across at least two categories, and scores from a model's own launch post are never used. A new model is listed straight away and ranked once independent publishers have measured it, usually within days to weeks.
- Which AI labs are due to release next?
- Going by release cadence alone, not announcements: Liquid AI (between 13 Sept 2026 and 7 Nov 2026), Zhipu AI (between 4 Oct 2026 and 27 Oct 2026) and Xiaomi (between 6 Oct 2026 and 27 Oct 2026). The table below has every lab.