LFM2.5-24B-A2B added, LFM3-1.7B added (8 Nov 2026)
LFM2.5-24B-A2B and LFM3-1.7B join the board. 2 new models, 2 top-10 rank changes, 9 price changes, 201 new benchmark results. Daily digest for 8 Nov 2026.
BenchLeader refreshed 1685 models from 28 of 28 data feeds (benchmark publishers, aggregators and price lists). 2 new models, 2 top-10 rank changes, 9 price changes, 201 new or updated benchmark results across 28 data feeds.
New models
- LFM2.5-24B-A2B (Liquid AI) is now listed at $0.21/M blended; no independent results yet. (details)
- LFM3-1.7B (Liquid AI) is now listed at $0.11/M blended; no independent results yet. (details)
Movement in the top 10
- Claude Fable 5.1 (max) moved from #6 to #5 in the BenchLeader Index. (details)
- GPT-6.1 Sol (max) moved from #5 to #6 in the BenchLeader Index. (details)
Price changes
- Gemma 4 26B A4B input price moved from $0.13/M to $0.15/M (15% up). (details)
- Gemma 4 26B A4B output price moved from $0.40/M to $0.60/M (50% up). (details)
- Gemma 4 26B A4B cached input price moved from $0.050/M to $0.015/M (70% down). (details)
- Hy4 cached input price moved from $0.042/M to $0.021/M (50% down). (details)
- Gemma 4 26B A4B (thinking) cached input price moved from $0.050/M to $0.015/M (70% down). (details)
- Gemma 4 26B A4B (no reasoning) cached input price moved from $0.050/M to $0.015/M (70% down). (details)
- Gemma 4 26B A4B (minimal) input price moved from $0.13/M to $0.15/M (15% up). (details)
- Gemma 4 26B A4B (minimal) output price moved from $0.40/M to $0.60/M (50% up). (details)
- Gemma 4 26B A4B (minimal) cached input price moved from $0.050/M to $0.015/M (70% down). (details)
New benchmark results
- Arena Search: 22 new results, including Claude Opus 4.6, GPT-5.5, o3… (board)
- Arena Document: 3 new results, including Claude Fable 5, Gemini 3.7 Flash, Gemini 3.8 Flash (board)
- Arena WebDev: 3 new results, including Inkling 2 Small, MiMo-V2.5, DeepSeek V3.2 (board)
- LiveCodeBench: 3 new results, including MiMo-V2.5, DeepSeek V3.2, DeepSeek V3 0324 (board)
- MMLU-Pro: 3 new results, including MiMo-V2.5, DeepSeek V3.2, DeepSeek V3 0324 (board)
- LegalBench: 3 new results, including MiMo-V2.5, DeepSeek V3.2, DeepSeek V3 0324 (board)
- CorpFin: 3 new results, including MiMo-V2.5, DeepSeek V3.2, DeepSeek V3 0324 (board)
- GPQA Diamond (Vals): 3 new results, including MiMo-V2.5, DeepSeek V3.2, DeepSeek V3 0324 (board)
- SciCode: 2 new results, including MiMo-V2.5, DeepSeek V3 0324 (board)
- ProofBench: 2 new results, including MiMo-V2.5, DeepSeek V3.2 (board)
- Arena Text: 2 new results, including MiMo-V2.5, DeepSeek V3.1 (board)
- Arena Hard Prompts: 2 new results, including MiMo-V2.5, DeepSeek V3.1 (board)
Score revisions
- Claude Opus 5.5 (max) on Coding-agent share: 55.9% → 56.7%. (details)
- Gemini 4 Carbon (high) on Arena Coding: 1573 → 1568. (details)
- GPT-6 Astra (max) on Arena Coding: 1542 → 1548. (details)
- Claude Fable 5.1 (max) on Arena Document: 1493 → 1499. (details)
- GPT-6.1 Sol (max) on Arena WebDev: 1751 → 1742. (details)
- GPT-6.1 Sol (max) on Arena Maths: 1509 → 1504. (details)
- GPT-6.1 Sol (max) on Arena Multi-turn: 1479 → 1487. (details)
- Claude Opus 5.5 (xhigh) on Coding-agent share: 55.9% → 56.7%. (details)
- Gemini 4 Argon (high) on Arena Creative Writing: 1517 → 1525. (details)
- Claude Opus 5.5 (high) on Arena Maths: 1512 → 1517. (details)
- Claude Opus 5.5 (high) on Coding-agent share: 55.9% → 56.7%. (details)
- GPT-6.1 Terra (max) on Arena WebDev: 1673 → 1680. (details)
- Claude Fable 5.1 (high) on Arena Search: 1233 → 1227. (details)
- Claude Opus 5.5 on Coding-agent share: 55.9% → 56.7%. (details)
- GPT-6 Sol (max) on Arena Creative Writing: 1435 → 1429. (details)
- Claude Sonnet 5.5 (xhigh) on Arena Hard Prompts: 1505 → 1499. (details)
- Claude Sonnet 5.5 (xhigh) on Arena Coding: 1529 → 1521. (details)
- Claude Sonnet 5.5 (xhigh) on Arena Maths: 1508 → 1488. (details)
- Claude Sonnet 5.5 (xhigh) on Arena Creative Writing: 1468 → 1463. (details)
- Claude Sonnet 5.5 (xhigh) on Arena Instruction Following: 1486 → 1481. (details)
- …and 51 more.
Speed changes
- GPT-6 Astra time to first answer changed from 55.98 s to 77.47 s. (details)
- GPT-5.6 Sol time to first token changed from 1.31 s to 3.29 s. (details)
- GPT-5.3-Codex time to first token changed from 1.75 s to 2.72 s. (details)
- Qwen3.7 time to first token changed from 0.74 s to 1.09 s. (details)
- GLM-5 time to first answer changed from 105.13 s to 56.97 s. (details)
- GLM-5 thinking time changed from 102.47 s to 54.16 s. (details)
- GLM 5.2 time to first answer changed from 44.32 s to 19.50 s. (details)
- GLM 5.2 thinking time changed from 40.86 s to 17.32 s. (details)
- GLM 5.2 time to first token changed from 0.44 s to 0.59 s. (details)
- Claude Opus 4.7 time to first answer changed from 52.30 s to 31.75 s. (details)
- Qwen3.6 Plus time to first answer changed from 73.02 s to 46.77 s. (details)
- Qwen3.6 Plus thinking time changed from 70.51 s to 44.07 s. (details)
- GPT-5.4 time to first answer changed from 11.22 s to 3.85 s. (details)
- Kimi K2.5 time to first answer changed from 62.35 s to 35.76 s. (details)
- Kimi K2.5 thinking time changed from 59.11 s to 31.74 s. (details)
- …and 130 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.