Source-labeled benchmark reference

AI model and company signals, clearly sourced.

CyberOGZ organizes model benchmarks, company data, market signals and metadata in one place so readers can compare sources faster and spot what still needs verification before making decisions.

Admin database Updated Sep 14, 2026 Company radar
Use every row as a source-labeled signal. Model scores shown here come from labeled external sources; CyberOGZ does not calculate or override the score. See how ›
AI company radar

Companies shaping the AI stack.

Public market signals, private AI lab profiles, and provider news are source-labeled when API-synced.

23Public 5Private labs 28API synced
Tencent Holdings Ltd logo TE

Tencent Holdings Ltd

TCEHY | Communication Services
Public

Cloud AI, gaming, social platforms and model ecosystem

HONG KONG EXCHANGES AND CLEARING LTDExchange 29Provider news China / AsiaRegion
+1.21%Move
Latest signal Tencent's Enflame Stake Gains Nearly $3 Billion in One Day
Provider data: Finnhub Synced Sep 14 Source
Apple Inc logo AP

Apple Inc

AAPL | Technology
Public

On-device AI, silicon, ecosystem distribution

Market signal $333.08 +0.24% today
NASDAQ NMS - GLOBAL MARKETExchange 249Provider news Global / USRegion
$333.08Price +0.24%Move $4.8TMarket cap
Latest signal X Corp and SpaceXAI drop antitrust lawsuit against Apple
Provider data: Finnhub Synced Sep 14 Source
Meta Platforms Inc logo ME

Meta Platforms Inc

META | Communication Services
Public

Llama models, recommendation systems and AI products

Market signal $665.6 +2.71% today
NASDAQ NMS - GLOBAL MARKETExchange 250Provider news Global / USRegion
$665.6Price +2.71%Move $1.7TMarket cap
Latest signal Prediction: Meta Stock Has More Upside Than Its Valuation Suggests
Provider data: Finnhub Synced Sep 14 Source
Broadcom Inc logo BR

Broadcom Inc

AVGO | Technology
Public

Networking silicon and custom AI accelerator supply chain

Market signal $344.72 -4.77% today
NASDAQ NMS - GLOBAL MARKETExchange 250Provider news Global / USRegion
$344.72Price -4.77%Move $1.7TMarket cap
Latest signal Vertiv vs. Broadcom: Which AI Infrastructure Stock Has More Upside?
Provider data: Finnhub Synced Sep 14 Source
Siemens AG logo SI

Siemens AG

SIEGY | Industrials
Public

Industrial AI, automation software and digital twin systems

XETRAExchange 24Provider news EuropeRegion
+2.51%Move
Latest signal Siemens AG: Riding The AI Investment Boom
Provider data: Finnhub Synced Sep 14 Source
Lenovo Group Ltd logo LE

Lenovo Group Ltd

LNVGY | Technology
Public

AI PCs, edge devices and enterprise hardware distribution

Market signal $83.04 +2.81% today
HONG KONG EXCHANGES AND CLEARING LTDExchange 27Provider news China / AsiaRegion
$83.04Price +2.81%Move $400.2BMarket cap
Latest signal Why Lenovo Group Ltd. (LNVGY) is a Top Dividend Stock for Your Portfolio
Provider data: Finnhub Synced Sep 14 Source
Perplexity logo PE

Perplexity

Private | Artificial Intelligence
Private

Perplexity AI unlocks the power of knowledge with information discovery and sharing.

Global / USRegion
Provider data: Brandfetch Logo + Brandfetch Brand API Synced Sep 14 Source
Mistral AI logo MI

Mistral AI

Private | Artificial Intelligence
Private

The most powerful AI platform for enterprises. Customize, fine-tune, and deploy AI assistants, autonomous agents, and multimodal AI with open models.

20Provider news EuropeRegion
Latest signal Mistral reportedly seeking $3 . 5B funding round amid physics AI push
Provider data: Brandfetch Logo + Brandfetch Brand API Synced Sep 14 Source
xAI logo XA

xAI

Private | Artificial Intelligence
Private

xAI is a company working on building artificial intelligence to accelerate human scientific discovery. We are guided by our mission to advance our collective understanding of the universe.

Global / USRegion
Provider data: Brandfetch Logo + Brandfetch Brand API Synced Sep 14 Source
Explore benchmark insights Source rankings, available benchmark breakdowns and release dates
Top performers

Top 10 Artificial Analysis LLM Intelligence rows

US / Global Europe China & Asia
  1. #1 Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback) Anthropic 53.4
  2. #2 Claude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback) Anthropic 53.2
  3. #3 GPT-6 Astra (max) OpenAI 52.8
  4. #4 GPT-6 Astra (xhigh) OpenAI 52.5
  5. #5 Claude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback) Anthropic 51.2
  6. #6 GPT-6 Astra (high) OpenAI 51
  7. #7 Claude Opus 5 (Adaptive Reasoning, Max Effort) Anthropic 50.7
  8. #8 Claude Opus 5 (Adaptive Reasoning, Xhigh Effort) Anthropic 49.7
  9. #9 GPT-6 Astra (medium) OpenAI 49.7
  10. #10 Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) Anthropic 49.7
Humanity's Last Exam

Top 8 Hugging Face HLE submissions

Standard HLE model-card results from the CAIS/HLE leaderboard. Hugging Face currently marks these submissions as unverified; tool-assisted variants are excluded.

HLE % correct
  1. #1 GLM-5.3 zai-org · Unverified model card 62.5%
  2. #2 Kimi-K3 moonshotai · Unverified model card 56%
  3. #3 Hy4-preview tencent · Unverified model card 55.4%
  4. #4 GLM-5.2 zai-org · Unverified model card 54.7%
  5. #5 Hy3 tencent · Unverified model card 53.2%
  6. #6 MiMo-V2.5-Pro XiaomiMiMo · Unverified model card 48%
  7. #7 Inkling thinkingmachines · Unverified model card 46%
  8. #8 Qwen3.8-2.4T-A95B Qwen · Unverified model card 43.6%
Release race

2026 model releases - through September

Each dot marks a source-provided release date. Hover, focus or tap for the model name.

2026 · 50 releases
Jan Feb Mar Apr May Jun Jul Aug Sep
Anthropic
OpenAI
Meta
Z AI
SpaceXAI
Kimi
Google
Region
Type
Model Score Coding Speed Input Output Source
Anthropic Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback) General
53.4 source-backed 81.6 65.05 tok/s $10 $50 Artificial Analysis
Anthropic Claude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback) General
53.2 source-backed 80.7 59.67 tok/s $10 $50 Artificial Analysis
OpenAI GPT-6 Astra (max) General
52.8 source-backed 76.9 58.05 tok/s $10 $50 Artificial Analysis
OpenAI GPT-6 Astra (xhigh) General
52.5 source-backed 75.9 54.82 tok/s $10 $50 Artificial Analysis
Anthropic Claude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback) General
51.2 source-backed 79.1 54.92 tok/s $10 $50 Artificial Analysis
OpenAI GPT-6 Astra (high) General
51 source-backed 77.1 52.41 tok/s $10 $50 Artificial Analysis
Anthropic Claude Opus 5 (Adaptive Reasoning, Max Effort) General
50.7 source-backed 78 51.7 tok/s $5 $25 Artificial Analysis
Anthropic Claude Opus 5 (Adaptive Reasoning, Xhigh Effort) General
49.7 source-backed 77 53.09 tok/s $5 $25 Artificial Analysis
OpenAI GPT-6 Astra (medium) General
49.7 source-backed 76.7 50.76 tok/s $10 $50 Artificial Analysis
Anthropic Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) General
49.7 source-backed 76.5 60.63 tok/s $10 $50 Artificial Analysis
Anthropic Claude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback) General
49.1 source-backed 77.1 53.62 tok/s $10 $50 Artificial Analysis
Anthropic Claude Opus 5 (Adaptive Reasoning, High Effort) General
48.2 source-backed 76.5 50.88 tok/s $5 $25 Artificial Analysis
Meta Muse Spark 1.3 (max) General
48.2 source-backed 75.8 229.98 tok/s $1.25 $4.25 Artificial Analysis
OpenAI GPT-5.6 Sol (max) General
47.1 source-backed 77.4 58.78 tok/s $4 $20 Artificial Analysis
Anthropic Claude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback) General
47 source-backed 75.2 53 tok/s $10 $50 Artificial Analysis
OpenAI GPT-6 Astra (low) General
46 source-backed 75.7 51.24 tok/s $10 $50 Artificial Analysis
Meta Muse Spark 1.3 (xhigh) General
45.2 source-backed 76.5 216.77 tok/s $1.25 $4.25 Artificial Analysis
Anthropic Claude Opus 5 (Adaptive Reasoning, Medium Effort) General
45.1 source-backed 74.3 51.23 tok/s $5 $25 Artificial Analysis
Z AI GLM-5.3 (max) General
China
44.9 source-backed 74.8 72.14 tok/s $1.4 $4.4 Artificial Analysis
SpaceXAI Grok 4.6 (high) General
44.4 source-backed 76.8 56.68 tok/s $2 $6 Artificial Analysis
SpaceXAI Grok 4.6 (xhigh) General
44.3 source-backed 75.9 54.63 tok/s $2 $6 Artificial Analysis
OpenAI GPT-5.6 Sol (xhigh) General
44.1 source-backed 78.3 60.05 tok/s $4 $20 Artificial Analysis
Kimi Kimi K3 (max) General
China
43.8 source-backed 76.2 35.7 tok/s $3 $15 Artificial Analysis
SpaceXAI Grok 4.6 (medium) General
43 source-backed 74.4 52.37 tok/s $2 $6 Artificial Analysis
OpenAI GPT-5.6 Sol (high) General
42.5 source-backed 77.2 57.1 tok/s $4 $20 Artificial Analysis
OpenAI GPT-5.6 Terra (max) General
42.3 source-backed 76.7 113.28 tok/s $2 $12 Artificial Analysis
Anthropic Claude Opus 4.8 (Adaptive Reasoning, Max Effort) General
42 source-backed 74.3 56.11 tok/s $5 $25 Artificial Analysis
Z AI GLM-5.3-Flash General
China
41.9 source-backed 71.5 106.97 tok/s $0.15 $0.5 Artificial Analysis
Google Gemini 3.8 Flash (high) General
41.2 source-backed 76.3 277.52 tok/s $0.75 $3.75 Artificial Analysis
Anthropic Claude Opus 4.7 (Adaptive Reasoning, Max Effort) General
40.7 source-backed 73.6 47.1 tok/s $5 $25 Artificial Analysis
Alibaba Qwen3.8 Max General
China
40.3 source-backed 71.8 40.69 tok/s $2 $6 Artificial Analysis
Google Gemini 3.8 Flash (medium) General
40 source-backed 74.1 n/a $0.75 $3.75 Artificial Analysis
Alibaba Qwen3.8 2.4T A95B General
China
40 source-backed 71.9 40.29 tok/s $2 $6 Artificial Analysis
Alibaba Qwen3.8-Flash-Next General
China
39.9 source-backed 73.1 51.2 tok/s $0.15 $0.47 Artificial Analysis
Meta Muse Spark 1.2 (xhigh) General
39.8 source-backed 72.2 177.8 tok/s $1.25 $4.25 Artificial Analysis
Anthropic Claude Opus 5 (Adaptive Reasoning, Low Effort) General
39.8 source-backed 66.9 53.02 tok/s $5 $25 Artificial Analysis
Google Gemini 3.7 Flash (medium) General
39.6 source-backed 71.5 278.55 tok/s $0.75 $3.75 Artificial Analysis
OpenAI GPT-5.6 Sol (medium) General
39.5 source-backed 76.3 59.69 tok/s $4 $20 Artificial Analysis
DeepSeek DeepSeek V4.1 Flash (Reasoning, Max Effort) General
China
39.5 source-backed n/a 214.39 tok/s $0.3 $1.2 Artificial Analysis
Google Gemini 3.7 Flash (high) General
39.4 source-backed 76.1 287.8 tok/s $0.75 $3.75 Artificial Analysis
SpaceXAI Grok 4.5 (high) General
39.1 source-backed 72.4 59.56 tok/s $2 $6 Artificial Analysis
OpenAI GPT-5.4 (xhigh) General
39 source-backed 71.1 145.2 tok/s $2.5 $15 Artificial Analysis
OpenAI GPT-5.5 (xhigh) General
38.6 source-backed 74.9 93.5 tok/s $5 $30 Artificial Analysis
Anthropic Claude Sonnet 5 (Adaptive Reasoning, Max Effort) General
38.4 source-backed 71.5 75.65 tok/s $2 $10 Artificial Analysis
OpenAI GPT-5.6 Terra (xhigh) General
38.2 source-backed 70.6 102.19 tok/s $2 $12 Artificial Analysis
OpenAI GPT-5.6 Luna (max) General
37.5 source-backed 71.4 119.66 tok/s $0.2 $1.2 Artificial Analysis
OpenAI GPT-5.5 (high) General
37.3 source-backed 71.6 95 tok/s $5 $30 Artificial Analysis
Google Gemini 3.7 Flash (low) General
36.9 source-backed 71 297.93 tok/s $0.75 $3.75 Artificial Analysis
DeepSeek DeepSeek V4 Pro 0813 (Reasoning, Max Effort) General
China
36.3 source-backed 68.8 82.11 tok/s $1.32 $3.96 Artificial Analysis
Sapiens AI Agnes 3.0 Flash General
35.5 source-backed n/a 234.69 tok/s $0.05 $0.15 Artificial Analysis
How to read the index

What affects it

Score imported/admin value Coding source-specific coding metric Reasoning source-specific reasoning metric Speed throughput or latency signal Context provider/API metadata Price provider/API pricing metadata

The index gives more weight to broad model quality, then includes coding, reasoning, speed, context and price efficiency. Treat close numbers as directional rather than absolute.

Overall is not a claim that one model is universally better than another. It is a compact reading aid for CyberOGZ readers, based on imported benchmark, speed, context and pricing signals at the time the data was added.

Use it to scan the field quickly, then compare the specific columns and source notes for the task you care about. A lower-index model can still be the better choice for a specific workflow.

Data sources and confidence

The benchmark feed can come from API-backed sources, imported JSON, and admin-reviewed manual rows. Each row has a source and confidence value. Higher confidence means the row is based on a clearer external source or a cleaner imported feed.

If a model has only metadata but no trusted benchmark data, it should stay inactive or metadata-only in admin. That prevents unscored models from polluting the benchmark lists.

How to read this: Score and ranking come from the linked benchmark source. Speed is source-reported throughput, not a CyberOGZ rating. Missing API fields stay unavailable.