Kronos TeknolojiResearch index · v1.3.0
Kronos AI Index
Open Frontier Delta measures the LiveBench percentage-point difference between a model and the highest model LiveBench designates open-weight in each of seven categories. It does not claim to measure universal intelligence.
Last updated: Sep 25, 2026, 2:12 PM UTC
Last LiveBench check: Sep 27, 2026, 12:01 PM UTC
Benchmark release: LiveBench 2026-06-25
Data status: Stale
Data source: LiveBench
Read the methodologyPrimary metric · methodology v1.3.0
Open Frontier Delta
Values are LiveBench percentage points above or below the LiveBench-designated open frontier.
| Claude Fable 5.1 Max Effort | Anthropic | -0.6 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
|---|---|---|---|---|---|---|---|---|
| Claude 5.5 Opus Thinking xHigh Effort · claude-opus-5-5-max-effort | Anthropic | -0.8 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Claude Fable 5 Max Effort | Anthropic | -1.0 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| GPT-6 Astra Max Effort | OpenAI | -1.8 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Claude 5.5 Opus Thinking xHigh Effort | Anthropic | -1.9 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Muse Spark 1.3 xHigh Effort | Meta | -2.4 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| DeepSeek V4.1 Flash Max Effort · 2026-09-10 | DeepSeek | -2.9 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| GPT-5.6 Sol Max Effort | OpenAI | -3.0 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| GPT-5.5 · xhigh | OpenAI | -3.8 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Claude 5 Opus Thinking xHigh Effort · claude-opus-5-max-effort | Anthropic | -3.9 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Smaug Agentic | Abacus.AI | -4.5 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| GPT-6 Sol Max Effort | OpenAI | -4.8 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Kimi K3 | Moonshot AI | -4.8 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Gemini 3.7 Flash High | -5.2 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 | |
| Qwen 3.8 Max | Alibaba | -5.5 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Grok 4.6 | xAI | -6.0 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| GPT-5.4 · xhigh | OpenAI | -6.0 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Muse Spark 1.2 xHigh Effort | Meta | -6.1 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| GPT-5.6 Terra Max Effort | OpenAI | -6.1 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| DeepSeek V4 Pro 0813 | DeepSeek | -6.6 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Smaug Flash | Abacus.AI | -6.6 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Grok 4.7 xHigh | xAI | -6.6 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Gemini 3.1 Pro Preview High | -7.1 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 | |
| Smaug Mini | Abacus.AI | -7.1 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| DeepSeek V4 Flash Vision Exp | DeepSeek | -7.2 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Claude 4.7 Opus Thinking High Effort · claude-opus-4-7-xhigh-effort | Anthropic | -7.5 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Claude 4.8 Opus Thinking Max Effort · 2026-04-16 | Anthropic | -7.8 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Qwen 3.8 Flash Next | Alibaba | -7.8 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| GLM-5.3 | Z.AI | -7.9 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Union Alpha | Stealth | -7.9 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Claude Sonnet 5 xHigh Effort | Anthropic | -8.0 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Gemini 3.8 Flash High | -8.2 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 | |
| Grok 4.5 | xAI | -8.2 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Muse Spark 1.1 xHigh Effort | Meta | -8.7 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Qwen3.8 27B | Alibaba | -8.7 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Gemini 3.5 Flash · high | -9.4 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 | |
| GPT-5.2 · high | OpenAI | -9.4 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Claude 4.6 Opus Thinking High Effort · 2026-02-05 | Anthropic | -9.5 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| DeepSeek V4 Flash 0731 | DeepSeek | -9.8 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| GPT-5.2 Codex · 2026-01-14 | OpenAI | -10.0 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Gemini 3.6 Flash High | -10.4 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 | |
| GPT-5.6 Luna Max Effort | OpenAI | -10.4 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| GLM-5.2 | Z.AI | -10.8 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Qwen 3.7 Max | Alibaba | -10.9 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Claude 4.6 Sonnet Thinking High Effort · claude-sonnet-4-6-thinking-auto-medium-effort | Anthropic | -11.0 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Claude 4.5 Opus Thinking · high-effort | Anthropic | -11.4 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| GPT-6 Luna Max Effort | OpenAI | -12.0 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Inkling xHigh Effort | Thinking Machines | -12.1 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| GLM-5.3 Flash | Z.AI | -12.4 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| deepseek-v4-pro | Unknown organization | -12.4 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Kimi K2.6 Thinking | Moonshot AI | -13.5 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| GPT-5.4 Nano · xhigh | OpenAI | -14.4 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| ox-alpha-max | Stealth | -14.8 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Qwen 3.6 Plus · 2026-04-02 | Alibaba | -15.1 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Kimi K2.7 Code | Moonshot AI | -15.6 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Grok Build 0.1 | xAI | -16.2 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Nemotron 3 Ultra 550B A55B | NVIDIA | -16.6 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Minimax M3 | Minimax | -16.7 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| GPT-5.4 Mini · xhigh | OpenAI | -17.6 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| deepseek-v4-flash | Unknown organization | -18.5 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Qwen 3.6 27B | Alibaba | -20.0 pp | 100% | 100 | LiveBench-designated open weight | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
| Gemini 3.5 Flash-Lite High | -20.1 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 | |
| Grok 4.3 | xAI | -21.8 pp | 100% | 100 | Not designated as open-weight by LiveBench | Not providedUSD per 1M tokens | Not provided | Sep 25, 2026 |
What the open frontier means
For each category, the ingestion reads LiveBench's openweight flag from the release-pinned model metadata and automatically selects the highest-scoring flagged model. A different model can lead each category. This source classification is not a separate Kronos verification of the model's license. The headline is the equal mean of all seven category deltas and is withheld if any category lacks a flagged baseline or comparable score.
AGI Capability Beta
Not scoredThere is not enough structured evidence for an honest composite score. A future beta may study breadth, depth, reliability, autonomy, adaptation, and robustness; each currently remains Insufficient evidence.
- BreadthInsufficient evidence
- DepthInsufficient evidence
- ReliabilityInsufficient evidence
- AutonomyInsufficient evidence
- AdaptationInsufficient evidence
- RobustnessInsufficient evidence
Snapshot history is versioned. New evidence appears only after an explicit calculation; it does not rewrite previously published scores.
Based on public LiveBench release data. Kronos applies the source's open-weight flag to calculate category deltas; this index is not affiliated with or endorsed by LiveBench.