all the models — AI benchmark observatory

Your AI Benchmark Observatory

all the models

Independent signal across paid and open-weight models — GPQA, SWE-bench, Arena Elo, and more, kept current from live sources.

AI Race

Trailing twelve months, week by week. Each lab’s line is its highest-scoring closed GPQA model released by that week.

.75.80.85.90.951.0Sep 29Nov 24Jan 19Mar 16May 11Jul 6Aug 31Sep 21
AnthropicGoogleOpenAIxAIMetaByteDanceAlibaba Cloud
  1. 1OpenAI0.960
    GPT-6 Astra
  2. 2Anthropic0.946
    Claude Mythos Preview
  3. 3Google0.943
    Gemini 3.1 Pro
  4. 4xAI0.930
    Grok 4.5
  5. 5Alibaba Cloud0.924
    Qwen3.7 Max
  6. 6Meta0.895
    Muse Spark
  7. 7ByteDance0.889
    Seed 2.0 Pro

Weeks in the lead

Same trailing twelve months · 52 weeks

  1. OpenAI21
  2. Anthropic17
  3. Google14
  4. Alibaba Cloud0
  5. ByteDance0
  6. Meta0
  7. xAI0
WeekLeaderAnthropicGoogleOpenAIxAIMetaByteDanceAlibaba Cloud
Sep 21OpenAI0.9460.9430.9600.9300.8950.8890.924
Sep 14OpenAI0.9460.9430.9600.9300.8950.8890.924
Sep 7OpenAI0.9460.9430.9600.9300.8950.8890.924
Aug 31OpenAI0.9460.9430.9600.9300.8950.8890.924
Aug 24Anthropic0.9460.9430.9460.9300.8950.8890.924
Aug 17Anthropic0.9460.9430.9460.9300.8950.8890.924
Aug 10Anthropic0.9460.9430.9460.9300.8950.8890.924
Aug 3Anthropic0.9460.9430.9460.9300.8950.8890.924
Jul 27Anthropic0.9460.9430.9460.9300.8950.8890.924
Jul 20Anthropic0.9460.9430.9460.9300.8950.8890.924
Jul 13Anthropic0.9460.9430.9460.9300.8950.8890.924
Jul 6Anthropic0.9460.9430.9460.8840.8950.8890.924
Jun 29Anthropic0.9460.9430.9360.8840.8950.8890.924
Jun 22Anthropic0.9460.9430.9360.8840.8950.8890.924
Jun 15Anthropic0.9460.9430.9360.8840.8950.8890.924
Jun 8Anthropic0.9460.9430.9360.8840.8950.8890.924
Jun 1Anthropic0.9460.9430.9360.8840.8950.8890.924
May 25Anthropic0.9460.9430.9360.8840.8950.8890.924
May 18Anthropic0.9460.9430.9360.8840.8950.8890.924
May 11Anthropic0.9460.9430.9360.8840.8950.8890.904
May 4Anthropic0.9460.9430.9360.8840.8950.8890.904
Apr 27Google0.9420.9430.9360.8750.8950.8890.904
Apr 20Google0.9420.9430.9360.8750.8950.8890.904
Apr 13Google0.9420.9430.9320.8750.8950.8890.904
Apr 6Google0.9130.9430.9320.8750.8950.8890.904
Mar 30Google0.9130.9430.9320.8750.8890.904
Mar 23Google0.9130.9430.9320.8750.8890.874
Mar 16Google0.9130.9430.9320.8750.8890.874
Mar 9Google0.9130.9430.9320.8750.8890.874
Mar 2Google0.9130.9430.9320.8750.8890.874
Feb 23Google0.9130.9430.9320.8750.8890.874
Feb 16Google0.9130.9430.9320.8750.8890.874
Feb 9OpenAI0.9130.9190.9320.8750.8890.874
Feb 2OpenAI0.9130.9190.9320.8750.620
Jan 26OpenAI0.8700.9190.9320.8750.620
Jan 19OpenAI0.8700.9190.9320.8750.620
Jan 12OpenAI0.8700.9190.9320.8750.620
Jan 5OpenAI0.8700.9190.9320.8750.620
Dec 29OpenAI0.8700.9190.9320.8750.620
Dec 22OpenAI0.8700.9190.9320.8750.620
Dec 15OpenAI0.8700.9190.9320.8750.620
Dec 8OpenAI0.8700.9190.9320.875
Dec 1Google0.8700.9190.8810.875
Nov 24Google0.8700.9190.8810.875
Nov 17Google0.8480.9190.8810.875
Nov 10OpenAI0.8480.8640.8810.875
Nov 3OpenAI0.8480.8640.8810.875
Oct 27OpenAI0.8480.8640.8810.875
Oct 20OpenAI0.8480.8640.8810.875
Oct 13OpenAI0.8480.8640.8810.875
Oct 6OpenAI0.8480.8640.8810.875
Sep 29OpenAI0.8480.8640.8810.875

Live ranking plane

Sorted by GPQA when available. Multi-source columns stay transparent.

398 models · 727 benches · 2357 scores · synced 9/24/2026, 12:15:08 PM

#ModelGPQASWEMMLU-ProHLEArenaAA Idx$/M in
1GPT-6 AstraOpenAIclosed0.96010.00
2Claude Mythos PreviewAnthropicclosed0.9460.9390.647
3GPT-5.6 SolOpenAIclosed0.9465.00
4Gemini 3.1 ProGoogleclosed0.9430.8060.51414872.00
5Claude Opus 4.7Anthropicclosed0.9420.8760.54714945.00
6Claude Opus 4.8Anthropicclosed0.9360.8860.57914735.00
7GPT-5.5OpenAIclosed0.9360.52214765.00
8Kimi K3Moonshot AIopen0.9350.5602.85
9GPT-5.2 ProOpenAIclosed0.9320.366
10Grok 4.5xAIclosed0.93014692.00
11GPT-5.6 TerraOpenAIclosed0.9292.00
12GPT-5.4OpenAIclosed0.9280.39814662.50
13Qwen3.8 MaxAlibaba Cloud / Qwen Teamopen0.9260.43614811.65
14GPT-5.2OpenAIclosed0.9240.8000.3451.75
15Qwen3.7 MaxAlibaba Cloud / Qwen Teamclosed0.9240.8040.8960.4141.25
16GPT-5.6 LunaOpenAIclosed0.9230.20
17Hy4 previewTencentopen0.923
18Gemini 3 ProGoogleclosed0.9190.7620.4581485
19Qwen3.8 FlashAlibaba Cloud / Qwen Teamclosed0.9170.3590.15
20Qwen3.8-Flash-NextAlibaba Cloud / Qwen Teamopen0.9170.359
21Claude Opus 4.6Anthropicclosed0.9130.8080.53114975.00
22GLM-5.2Zhipu AIopen0.9120.5470.75
23DeepSeek-V4.1-FlashDeepSeekopen0.9090.3680.22
24Kimi K2.6Moonshot AIopen0.9050.8020.3640.75
25Gemini 3 FlashGoogleclosed0.9040.7800.43514740.50
26Hy3Tencentopen0.9040.7800.14
27Qwen3.6 PlusAlibaba Cloud / Qwen Teamclosed0.9040.7880.8850.2880.50
28Qwen3.7-PlusAlibaba Cloud / Qwen Teamclosed0.9030.7770.8850.347
29DeepSeek-V4-Pro-MaxDeepSeekopen0.9010.8060.8750.4821.30
30Claude Sonnet 4.6Anthropicclosed0.8990.7960.49014733.00
31Inkling-SmallThinking Machines Labopen0.8950.8020.3160.30
32Muse SparkMetaclosed0.8950.7740.5841488
33Qwen3.8-27BAlibaba Cloud / Qwen Teamopen0.8920.3080.40
34Solar Pro 4Upstageclosed0.8900.7060.8630.30
35Seed 2.0 ProByteDanceclosed0.8890.7650.50
36Grok-4 HeavyxAIclosed0.8840.507
37Qwen3.5-397B-A17BAlibaba Cloud / Qwen Teamopen0.8840.7640.8780.2870.45
38DeepSeek-V4-Flash-MaxDeepSeekopen0.8810.7900.8620.4510.09
39GPT-5 MediumOpenAIclosed0.881
40GPT-5.1OpenAIclosed0.8810.7631.25