Intelligence index
The frontier this week
Composite quality, coding, price, and speed for models you can actually call. Scores are collated in the spirit of Artificial Analysis v4.1–v4.2, September 2026 — a reported leaderboard, not AiSignal’s own benchmark. Tap up to three rows to compare.
| Model | Context | ||||
|---|---|---|---|---|---|
Claude Fable 5.1 Anthropic · closed | 56.8 | 79.1 | $20.0 | 70 | 1.0M |
GPT-6 Astra OpenAI · closed | 54.7 | 76.2 | $50.0 | 63 | 1.0M |
Claude Opus 5 Anthropic · closed | 54.1 | 78.0 | $10.0 | 52 | 1.0M |
Claude Fable 5 Anthropic · closed | 53.2 | 76.5 | $20.0 | 62 | 1.0M |
Muse Spark 1.3 Meta · closed | 53.0 | 76.3 | $2.0 | 177 | 1.0M |
GPT-5.6 Sol OpenAI · closed | 51.3 | 77.2 | $8.0 | 107 | 1.0M |
Grok 4.6 xAI · closed | 50.6 | 76.8 | $3.0 | 57 | 500K |
Kimi K3 Moonshot · open weights | 50.2 | 76.2 | $5.1 | 276 | 1.0M |
GLM-5.3 Z.AI · open weights | 48.6 | 74.8 | $1.7 | 412 | 1.0M |
Gemini 3.8 Flash Google · closed | 47.1 | 76.3 | $1.5 | 346 | 1.0M |
Qwen3.8 Max Alibaba · open weights | 46.9 | 71.8 | $3.0 | 39 | 1.0M |
GPT-5.6 Terra OpenAI · closed | 46.8 | 75.1 | $4.5 | 133 | 1.0M |
Grok 4.5 xAI · closed | 45.5 | 72.4 | $3.0 | 50 | 500K |
Claude Sonnet 5 Anthropic · closed | 45.1 | 71.5 | $4.0 | 81 | 1.0M |
DeepSeek V4 Pro DeepSeek · open weights | 44.0 | 69.4 | $0.4 | 88 | 256K |
Not an official Artificial Analysis reprint. Independent collation for editorial use. Prices are list output rates and ignore cache discounts.