XBench

Home = our average · lab pages = their ranks

As of 2026-08-14

Artificial Analysis.

Their board. Their ranks.

Artificial Analysis Intelligence, Agentic, and Coding indices, plus cost per task and speed.

AA #1 is Opus 5

Open the lab’s original

Artificial Analysis

Copied from the AA current-status board (v4.1.1) and the Grok 4.6 AA article. No averaging.

1
Opus 5
Anthropic · closed
Anthropic’s flagship frontier model
63Lab59.2AA Agentic78AA Coding$2.34Lab53Lab
2
Fable 5
Anthropic · closed
New Claude 5 frontier tier
62Lab56.6AA Agentic76.5AA Coding$3.14Lab63Lab
3
GPT-5.6 Sol
OpenAI · closed
Top-tier flagship reasoning
61Lab57.8AA Agentic77.4AA Coding$1.23Lab62Lab
3
Grok 4.6
SpaceXAI · closed
Latest Grok frontier model
61Lab58.7AA Agentic76.8AA Coding$0.84Lab66Lab
5
Kimi K3
Moonshot AI · open
Frontier open-weight
60Lab54.3AA Agentic76.2AA Coding$0.84Lab39Lab
6
Qwen3.8 Max
Alibaba · closed
Hosted 2.4T flagship; weights promised
58Lab58.4AA Agentic71.8AA Coding$1.13Lab47Lab
6
Qwen3.8 open
Alibaba · open
Open-weight Qwen3.8 (AA #2 open)
58Lab58.4AA Agentic71.8AA Coding$1.09Lab52Lab
8
Spark 1.2
Meta · closed
Agent-native Meta model
57Lab49.3AA Agentic72.2AA Coding$0.40Lab
8
GPT-5.6 Terra
OpenAI · closed
Balanced GPT-5.6
57Lab50.2AA Agentic76.7AA Coding$0.51Lab115Lab
10
3.7 Flash
Google DeepMind · closed
Latest Flash workhorse
56Lab45.1AA Agentic76.1AA Coding$0.40Lab340Lab
10
Grok 4.5
SpaceXAI · closed
Frontier coding/agents
56Lab48.9AA Agentic72.5AA Coding$0.36Lab60Lab
10
Opus 4.8
Anthropic · closed
Frontier reasoning/coding
56Lab49.4AA Agentic74.3AA Coding$1.80Lab
13
Sonnet 5
Anthropic · closed
Frontier general-purpose/agentic
55Lab49.7AA Agentic71.5AA Coding$1.72Lab66Lab
14
GLM-5.2
Zhipu AI · open
Frontier/open-weight coding
53Lab45.7AA Agentic68.8AA Coding$0.32Lab108Lab
14
V4 Pro 0813
DeepSeek · open
Current DeepSeek-V4 open challenger
53Lab49.6AA Agentic68.8AA Coding$0.25Lab81Lab
16
V4 Flash
DeepSeek · open
Open price-performance pick
52Lab48.4AA Agentic69.1AA Coding$0.03Lab121Lab
16
GPT-5.6 Luna
OpenAI · closed
Cost-efficient GPT-5.6
52Lab46.9AA Agentic71.5AA Coding$0.05Lab152Lab
16
3.6 Flash
Google DeepMind · closed
Frontier-class fast reasoning
52Lab40.5AA Agentic69.2AA Coding$0.56Lab225Lab
19
3.1 Pro
Google DeepMind · closed
Frontier multimodal/reasoning
48Lab23.1AA Agentic68.8AA Coding$0.33Lab113Lab
20
V4 Pro
DeepSeek · open
Frontier open/low-cost challenger
45Lab$0.05Lab
20
MiniMax M3
MiniMax · open
Open-weight frontier-adjacent
45Lab36.1AA Agentic58.6AA Coding$0.14Lab86Lab
22
MiMo V2.5 Pro
Xiaomi · open
Open MIT, cheap per-task
43Lab29.5AA Agentic60.2AA Coding$0.03Lab46Lab
23
Inkling
Thinking Machines · open
New open lab, Apache-2.0
42Lab34.1AA Agentic52.1AA Coding$0.34Lab73Lab
23
Hy3
Tencent · open
Open Apache-2.0 coding/reasoning
42Lab31.4AA Agentic58.8AA Coding$0.04Lab64Lab
25
Qwen3.7 Max
Alibaba · closed
Frontier Chinese model
39Lab30.9AA Agentic66AA Coding$0.24Lab
26
Grok 4.3
SpaceXAI · closed
Frontier reasoning/agents
37Prelim24.2AA Agentic42.3AA Coding
27
LongCat 2.0
Meituan · open
Frontier open-weight agentic coding
34Prelim
28
GPT-5.5
OpenAI · closed
Frontier reasoning + agents
29Lab47.4AA Agentic74.9AA Coding
29
Doubao Seed
ByteDance · closed
Frontier reasoning
26Prelim
Opus 4.7
Anthropic · closed
Frontier coding/reasoning
46.3AA Agentic73.6AA Coding

30 models on this board · # is this page’s rank