Reaching Intelligence Index 30 Now Costs 86x Less Per Task Than When First Reached
ArtificialAnlys · x · 2026-10-21
Artificial Analysis measured cost per task at each Intelligence Index threshold (v4.3.2, prices as of Oct 20, 2026).
- Reaching index 20 is 123x cheaper than 13 months ago; reaching 40 is 33x cheaper than 5 months ago
- GPT-6 Luna (high) hits 30 at $0.03/task — 86x cheaper than Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort), which first reached 30 in Feb 2026 at $2.49/task
- Claude Haiku 5.5 (xhigh) is now the cheapest way to reach 40: scores 41 at $0.12/task, 5% below GPT-6.1 Sol (low)
- Reaching 50 is 18x cheaper than at first arrival: GPT-6.1 Sol (high) does it for $0.32/task
- The top measured score rose from 21 to 58 in a year, at 11x the cost: Claude Opus 5.5 (Max Effort) at $5.98/task
Takeaway: established intelligence tiers commoditize fast, while the frontier keeps getting more expensive.
More from Infra
- Corporate token spending growth collapses from 42% to 0.2% as hyperscaler ROIC falls to 20% — SumitGup · 2026-10-21
- Qualcomm plans worldwide performance-based layoffs as it chases $5B AI data center revenue — rohanpaul_ai · 2026-10-21
- vLLM at PyTorchCon 2026: KV cache and kernel optimizations across TPU, AMD, and NVIDIA — PyTorch · 2026-10-21
- antirez to port DeepSeek Infer kernels to DGX Spark's SM121 for DwarfStar — antirez · 2026-10-21
- Ternary-Bonsai-2-9B: 2-bit ternary vision-language model trends on Hugging Face for on-device use — prism-ml · 2026-10-21
- General Compute x Positron: GPUs Handle Prefill, Memory-First Chips Tackle Decode-Bound LLM Inference — cneuralnetwork · 2026-10-21