Changelog
What changed on BenchLeader each day: models ranked for the first time, new models, new results from 28 independent sources, moves in the BenchLeader Index, and price changes.
Updated every six hours; data as of 16 Nov 2026. Speed and rating drift are in each day's digest.
Get new models and rankings by emailThe last 14 days, from 3 Nov 2026. Every earlier day, back to 8 Sep 2026, is in the daily digests.
Prices changed for 3 models
- Qwen3.5-9Bthinkingcache read $0.063 → $0.137/M (+117%)
- Qwen3.5-9Bno reasoningcache read $0.063 → $0.137/M (+117%)
- Qwen3.5-9Bcache read $0.063 → $0.137/M (+117%)
Weekly review
Claude Fable 5.2 + 8 new AI models: 9 Nov–15 Nov9 new models incl. Claude Fable 5.2, Gemini 4 Hydrogen and Muse Ember, Gemma 4 31B up 4.4, 19 price cuts, 1414 new benchmark results: what moved on the BenchLeader Index this week.
New model listed
Nemotron 3.5 Bolt E4Bthinking
Vals AI published 317 new results
- Claude Opus 5.5maxIOI 95.1%
- Claude Opus 5.5maxVals Index 70.1
- Claude Opus 5.5maxBioMysteryBench 79.3%
- Claude Opus 5.5maxClaimBench 100.0%
- and 313 more · all in the digest
LiveBench published 200 new results
- Claude Fable 5.2xhighLiveBench 2.0 84.9%
- Claude Fable 5.2xhighLiveBench Long Agentic 82.9%
- Claude Fable 5.2xhighLiveBench 2.0 Reasoning 88.8%
- Claude Fable 5.2xhighLiveBench 2.0 Coding 79.2%
- and 196 more · all in the digest
Aider polyglot leaderboard published 25 new results
- GPT-5highAider Polyglot 88.0%
- GPT-5mediumAider Polyglot 86.7%
- o3-prohighAider Polyglot 84.9%
- Gemini 2.5 ProthinkingAider Polyglot 83.1%
- and 21 more · all in the digest
DuelLab GameBench 2 published 10 new results
- Claude Fable 5.2xhighGameBench 2 by reasoning setting 99.1
- Claude Fable 5.2mediumGameBench 2 by reasoning setting 94.0
- Claude Fable 5.2GameBench 2 Overall 95.0
- Claude Fable 5.2lowGameBench 2 by reasoning setting 91.8
- and 6 more · all in the digest
6 more sources published results · all in the digest
BenchLeader Index order changed
- Claude Fable 5.1max#9 → #7
- GPT-6 Astra#7 → #9
Prices changed for 38 models
- MiMo-V2.5-Procache read $0.2 → $0.004/M (-98%)
- MiMo-V2.5-Prothinkingcache read $0.2 → $0.004/M (-98%)
- MiMo-V2.5-Prono reasoningcache read $0.2 → $0.004/M (-98%)
- Nano Banana 2.1output $30 → $7.5/M (-75%)
- and 34 more · all in the digest
Artificial Analysis published 496 new results
- Claude Fable 5.1maxGDPval-AA v2.1 1765
- Claude Fable 5.1xhighGDPval-AA v2.1 1741
- Claude Opus 5maxGDPval-AA v2.1 1729
- Claude Opus 5xhighGDPval-AA v2.1 1695
- and 492 more · all in the digest
Epoch AI Benchmarking Hub published 3 new results
- Gemini 4 CarbonhighTerminal-Bench 64.5%
- Gemini 4 ArgonhighTerminal-Bench 55.5%
- Muse WonderxhighTerminal-Bench 47.0%
Design Arena published 2 new results
- Gemini 4 HydrogenDesign Arena Agentic 1265
- Hy3Design Arena Agentic 1156
BenchLeader Index order changed
- GPT-6.1 Solmax#9 → #8
- Claude Fable 5.1max#8 → #9
Prices changed for 16 models
- DeepSeek V4 Pro 0424maxinput $1.31 → $1.5/M (+15%)
- Muse Glimmerhighoutput $1.2 → $1.35/M (+13%)
- Nemotron 3 Ultra 550B A55Bthinkingoutput $2.7 → $2.4/M (-11%)
- Nemotron 3.5 Super VLinput $0.18 → $0.2/M (+11%)
- and 12 more · all in the digest
Artificial Analysis published 36 new results
- Gemini 4 ArgonmediumAA Intelligence Index v5.0.1 47.6
- Gemini 4 ArgonmediumHumanity's Last Exam (AA) 49.9%
- Gemini 4 ArgonmediumHLE-Diamond (AA) 42.5%
- Gemini 4 ArgonmediumAA-LCR v1.1 78.3%
- and 32 more · all in the digest
Epoch AI Benchmarking Hub published 19 new results
- Claude Fable 5.2maxTerminal-Bench 65.8%
- Kimi K3.1 PromaxTerminal-Bench 59.7%
- Gemini 4 HydrogenhighTerminal-Bench 56.4%
- GPT-6.1 TerramaxTerminal-Bench 53.3%
- and 15 more · all in the digest
Arena (LMArena) published 9 new results
- Gemini 4 HydrogenhighArena Text 1490
- Gemini 4 HydrogenhighArena Hard Prompts 1514
- Gemini 4 HydrogenhighArena Coding 1518
- Gemini 4 HydrogenhighArena WebDev 1704
- and 5 more · all in the digest
Scale AI SEAL published 4 new results
- Kimi K3.1 ProHiL-Bench 61.3%
- Gemini 4 CarbonHiL-Bench 50.3%
- Muse EmberFORTRESS 11.9%
- Inkling 2 SmallFORTRESS 16.2%
1 more source published results · all in the digest
Prices changed for 5 models
- DeepSeek V4 Prohighinput $0.66 → $1.32/M (+100%)
- DeepSeek V4 Prohighoutput $1.98 → $3.96/M (+100%)
- Ripple-65B Previewinput $0.2 → $0.1/M (-50%)
- Ripple-65B Previewoutput $0.6 → $0.3/M (-50%)
- and 1 more · all in the digest
New model listed
MathArena published 22 new results
- Gemini 3.5 FlashhighAIME 2026 95.0%
- Gemini 3.5 FlashhighHMMT February 2026 95.5%
- Gemini 3.5 FlashhighMathArena Apex 32.3%
- Kimi K3maxAIME 2026 96.7%
- and 18 more · all in the digest
Artificial Analysis published 11 new results
- Hy4MLCR 32.8%
- Qwen3.8-Flash-NextAA Intelligence Index v5.0.1 28.6
- Qwen3.8-Flash-NextFrontierCode 2.0 (AA) 27.1%
- Qwen3.8-Flash-NextTerminal-Bench-Science 0.1 (AA) 7.14%
- and 7 more · all in the digest
Kaggle Benchmarks published 10 new results
- GPT-5.6 SolFACTS Search 92.9%
- GPT-6.1 TerraFACTS Search 92.7%
- GPT-6 SolFACTS Search 92.7%
- Grok 4.6FACTS Search 87.9%
- and 6 more · all in the digest
Epoch AI Benchmarking Hub published 9 new results
- Claude Fable 5.1maxPostTrainBench 44.6%
- Claude Opus 5.5maxPostTrainBench 43.8%
- GPT-6 AstramaxPostTrainBench 41.9%
- Gemini 4 ArgonhighPostTrainBench 40.4%
- and 5 more · all in the digest
4 more sources published results · all in the digest
BenchLeader Index order changed
- GPT-6 Astramax#7 → #6
- GPT-6 Astra#8 → #7
- Claude Fable 5.1max#9 → #8
- GPT-6.1 Solmax#10 → #9
- and 2 more · all in the digest
Prices changed for 17 models
- Gemini 4 Argonhighprice $9.95 → $4.97/M (-50%)
- Gemini 4 Carbonhighinput $4 → $2/M (-50%)
- Gemini 4 Carbonhighoutput $20 → $10/M (-50%)
- Gemini 4 Carbonmediuminput $4 → $2/M (-50%)
- and 13 more · all in the digest
New model listed
Epoch AI Benchmarking Hub published 108 new results
- Claude Fable 5.2maxSciCode 67.7%
- Claude Fable 5.2maxCursorBench 60.0%
- Claude Fable 5.2maxFrontierSWE 69.9%
- Claude Fable 5.2xhighSciCode 65.6%
- and 104 more · all in the digest
ARC Prize published 9 new results
- Gemini 4 CarbonhighARC-AGI-1 97.5%
- Gemini 4 CarbonhighARC-AGI-2 92.1%
- Gemini 4 CarbonhighARC-AGI-3 96.9%
- Gemini 4 CarbonmediumARC-AGI-1 98.5%
- and 5 more · all in the digest
Artificial Analysis published 8 new results
- Gemini 4 CarbonhighMLCR 19.4%
- Hy4AA Cyber Index 58.0
- Hy4CWE-Bench (AA) 75.8%
- Hy4DeepSec-Bench (AA) 25.6%
- and 4 more · all in the digest
LiveBench published 8 new results
- Gemini 4 CarbonLiveBench 2.0 83.5%
- Gemini 4 CarbonLiveBench Long Agentic 72.7%
- Gemini 4 CarbonLiveBench 2.0 Reasoning 90.6%
- Gemini 4 CarbonLiveBench 2.0 Coding 91.0%
- and 4 more · all in the digest
5 more sources published results · all in the digest
BenchLeader Index order changed
- Claude Fable 5.2xhigh#4 → #2
- Gemini 4 Carbonhigh#5 → #3
- Claude Opus 5.5max#3 → #4
- Claude Fable 5.2high#2 → #5
- and 3 more · all in the digest
Prices changed for 5 models
- Qwen2.5 7B Instructoutput $0.2 → $0.05/M (-75%)
- Nemotron 3 Nano Omni 30B A3Bthinkinginput $0.5 → $0.2/M (-60%)
- Qwen2.5 7B Instructinput $0.1 → $0.05/M (-50%)
- Nemotron 3 Nano Omni 30B A3Bthinkingoutput $0.9 → $0.8/M (-11%)
- and 1 more · all in the digest
Epoch AI Benchmarking Hub published 29 new results
- Claude Fable 5.2maxGPQA Diamond 93.9%
- Claude Fable 5.2maxFrontierMath Tiers 1–3 95.1%
- Claude Fable 5.2maxFrontierMath Tier 4 100.0%
- Claude Fable 5.2maxOTIS Mock AIME 99.7%
- and 25 more · all in the digest
Artificial Analysis published 13 new results
- Claude Fable 5.2maxAA Progress Index v1 358
- Claude Fable 5.2maxMLCR 43.9%
- Inkling 2 SmallMLCR 20.6%
- Muse EmbermaxMLCR 18.3%
- and 9 more · all in the digest
Arena (LMArena) published 8 new results
- Kimi K3.1 PromaxArena Text 1483
- Kimi K3.1 PromaxArena Hard Prompts 1501
- Kimi K3.1 PromaxArena Coding 1526
- Kimi K3.1 PromaxArena Creative Writing 1457
- and 4 more · all in the digest
Vals AI published 2 new results
- Claude Fable 5.2maxVals RSI Index 37.9%
- Gemini 4 CarbonhighVals RSI Index 27.7%
2 more sources published results · all in the digest
BenchLeader Index order changed
- Claude Fable 5.2high#3 → #2
- Claude Opus 5.5max#2 → #3
Prices changed for 9 models
- MiMo-V2.5-Prono reasoningoutput $1.85 → $0.7/M (-62%)
- MiMo-V2.5-Prono reasoninginput $0.67 → $0.35/M (-48%)
- MiMo-V2.5input $0.18 → $0.11/M (-39%)
- MiMo-V2.5output $0.36 → $0.22/M (-39%)
- and 5 more · all in the digest
New model listed
New model listed
Arena (LMArena) published 8 new results
- Inkling 2 SmallArena Text 1412
- Inkling 2 SmallArena Hard Prompts 1438
- Inkling 2 SmallArena Coding 1486
- Inkling 2 SmallArena Agent 0.1
- and 4 more · all in the digest
Artificial Analysis published a new result
- gpt-oss-120bhighCritPt v2 hard subset (AA) 4.00%
BenchLeader Index order changed
- Claude Opus 5.5max#1 → #2
- Gemini 4 Carbonhigh#2 → #5
- GPT-6 Astra#3 → #6
- GPT-6 Astramax#4 → #7
- and 6 more · all in the digest
Prices changed for one model
- Muse Glimmerhighprice $0.161 → $0.17/M (+5%)
Weekly review
Kimi K3.1 Pro Together + 11 new AI models: 2 Nov–8 Nov12 new models incl. Kimi K3.1 Pro Together, Inkling 2 Small and LFM2.5-24B-A2B, DeepSeek V3.1 up 3.5, 20 price cuts, 314 new benchmark results: what moved on the BenchLeader Index this week.
Arena (LMArena) published 45 new results
- Claude Fable 5highArena Document 1500
- Claude Opus 4.6Arena Search 1222
- Gemini 3.8 FlashhighArena Document 1469
- Gemini 3.7 FlashhighArena Document 1468
- and 41 more · all in the digest
Vals AI published 39 new results
- MiMo-V2.5LiveCodeBench 81.5%
- MiMo-V2.5MMLU-Pro 82.9%
- MiMo-V2.5LegalBench 78.9%
- MiMo-V2.5CorpFin 59.9%
- and 35 more · all in the digest
Epoch AI Benchmarking Hub published 20 new results
- MiMo-V2.5SciCode 43.1%
- MiMo-V2.5ProofBench 16.0%
- MiMo-V2.5ALE-Bench 514
- DeepSeek V3.2Terminal-Bench 39.5%
- and 16 more · all in the digest
Artificial Analysis published 15 new results
- DeepSeek V3 0324AA Progress Index v1 195
- DeepSeek V3 0324GPQA Diamond (AA) 65.5%
- DeepSeek V3 0324Humanity's Last Exam (AA) 4.74%
- DeepSeek V3 0324IFBench 41.0%
- and 11 more · all in the digest
4 more sources published results · all in the digest
BenchLeader Index order changed
- Claude Fable 5.1max#6 → #5
- GPT-6.1 Solmax#5 → #6
Prices changed for 9 models
- Gemma 4 26B A4Bcache read $0.05 → $0.015/M (-70%)
- Gemma 4 26B A4Bthinkingcache read $0.05 → $0.015/M (-70%)
- Gemma 4 26B A4Bno reasoningcache read $0.05 → $0.015/M (-70%)
- Gemma 4 26B A4Bminimalcache read $0.05 → $0.015/M (-70%)
- and 5 more · all in the digest
New model listed
New model listed
Artificial Analysis published 13 new results
- GLM 5.2maxHLE-Diamond (AA) 17.7%
- GLM 5.2maxAA-Omniscience v1.1 0.6
- GLM 5.2maxAA-Omniscience v1.1: accuracy 21.1%
- GLM 5.2maxAA-Omniscience v1.1: non-hallucination 74.1%
- and 9 more · all in the digest
Kaggle Benchmarks published 2 new results
- Gemini 4 CarbonFACTS Search 94.4%
- Gemini 4 ArgonFACTS Search 89.9%
Scale AI SEAL published 2 new results
- Gemini 4 ArgonMCP Atlas 90.7%
- Claude Opus 5.5MCP Atlas 85.4%
BenchLeader Index order changed
- GPT-6.1 Solmax#6 → #5
- Claude Fable 5.1max#5 → #6
Prices changed for 7 models
- Nemotron 3 Super 120B A12Boutput $2.4 → $0.9/M (-62%)
- DeepSeek V3.1thinkinginput $0.58 → $0.43/M (-26%)
- DeepSeek V3.1thinkingoutput $1.68 → $1.34/M (-20%)
- Nemotron 3.5 Super VLinput $0.2 → $0.18/M (-10%)
- and 3 more · all in the digest
New model listed
Artificial Analysis published 63 new results
- Claude Opus 5maxAA Intelligence Index v5.0.1 44.2
- Claude Opus 5maxHLE-Diamond (AA) 44.8%
- Claude Opus 5maxAA-Omniscience v1.1 37.6
- Claude Opus 5maxCritPt v2 hard subset (AA) 50.0%
- and 59 more · all in the digest
Scale AI SEAL published 8 new results
- Beam 1.0 501B A23BFORTRESS 11.9%
- GPT-6 AstraFORTRESS 12.1%
- GLM 5.3 FlashFORTRESS 26.7%
- MiniMax M3.1 FlashPRBench Finance 45.6%
- and 4 more · all in the digest
Epoch AI Benchmarking Hub published 6 new results
- Doubao Seed EvolvinghighChess Puzzles 40.0%
- Doubao Seed 2.1 Turbo 260628highChess Puzzles 26.0%
- Doubao Seed 2.1 Pro 260628highChess Puzzles 21.0%
- Doubao Seed 2.0 Lite 260428highChess Puzzles 20.0%
- and 2 more · all in the digest
Kaggle Benchmarks published 5 new results
- Kimi K3FACTS Search 0.90%
- GLM 5.3FACTS Search 0.88%
- Claude Sonnet 5.5FACTS Search 0.87%
- Qwen3.8 MaxmaxFACTS Search 0.85%
- and 1 more · all in the digest
Prices changed for one model
- Gemini 4 Argonhighprice $7.68 → $9.95/M (+29%)
Artificial Analysis published 30 new results
- GPT-6 SolmaxAA Intelligence Index v5.0.1 44.1
- GPT-6 SolmaxAA-Omniscience v1.1 21.4
- GPT-6 SolmaxCritPt v2 hard subset (AA) 66.0%
- GPT-6 SolmaxAA-Omniscience v1.1: accuracy 51.1%
- and 26 more · all in the digest
DuelLab GameBench 2 published 7 new results
- Claude Fable 5.1xhighGameBench 2 by reasoning setting 94.1
- Claude Fable 5.1mediumGameBench 2 by reasoning setting 82.9
- Kimi K3.1 ProGameBench 2 Overall 78.2
- Claude Fable 5.1lowGameBench 2 by reasoning setting 84.8
- and 3 more · all in the digest
Vals AI published a new result
- Kimi K3.1 PromaxSREBench 27.1%
BenchLeader Index order changed
- Claude Fable 5.1max#6 → #5
- GPT-6.1 Solmax#5 → #6
Prices changed for 171 models
- Qwen-MT Plusoutput $0.775 → $7.37/M (+851%)
- Qwen-MT Plusinput $0.259 → $2.46/M (+850%)
- Qwen-VLmaxoutput $0.574 → $3.2/M (+457%)
- Qwen-Plusoutput $0.287 → $1.2/M (+318%)
- and 167 more · all in the digest
New model listed
ARC Prize published a new result
- DeepSeek V4.1 Flash 1017maxARC-AGI-3 5.80%
Kaggle Benchmarks published a new result
- gpt-oss-20bBrowseComp 1.26%
Artificial Analysis published a new result
- Apodex 1.1Terminal-Bench-Science 0.1 (AA) 0.48%
Arena (LMArena) published a new result
- Spark X2.6Arena Agent -10
Prices changed for 81 models
- GLM 4.5Vinput $0.6 → $0.299/M (-50%)
- GLM 4.5Voutput $1.8 → $0.896/M (-50%)
- GLM 4.6Voutput $0.9 → $0.448/M (-50%)
- GLM 4.6Vinput $0.3 → $0.149/M (-50%)
- and 77 more · all in the digest