Claude Fable 5 leads HLE (Humanity's Last Exam) in 2026 at 53.3%. Claude Opus 5 Max leads Intelligence Index at 60.7 (Xhigh 60.1, Fable 5 59.9). GPT-5.6 Sol max (HLE 47.2%, GPQA 94.1% co-leader), Claude Opus 4.8 (HLE 45.7%) compared. Ranked by GPQA Diamond, AIME, MATH-500, HLE. Free.
| # | Model | Score | Benchmarks | Input $/M | Output $/M | Speed | TTFT |
|---|---|---|---|---|---|---|---|
| 1 | 95 | 100 | $10.00 | $50.00 | 66 | 282.34s | |
| 2 | 94 | 99 | $10.00 | $50.00 | 69 | 111.68s | |
| 3 | 91 | 96 | $5.00 | $25.00 | 59 | 77.25s | |
| 4 | 91 | 95 | $5.00 | $25.00 | 55 | 32.99s | |
| 5 | 90 | 95 | $10.00 | $50.00 | 56 | 30.36s | |
| 6 | 90 | 94 | $10.00 | $50.00 | 70 | 104.62s | |
| 7 | 89 | 94 | $5.00 | $25.00 | 56 | 17.14s | |
| 8 | 89 | 92 | $1.25 | $4.25 | 186 | 42.55s | |
| 9 | Grok 4.6 (high) SpaceXAI | 88 | 93 | $2.00 | $6.00 | 61 | 47.23s |
| 10 | GPT-5.6 Sol (max) OpenAI | 88 | 93 | $4.00 | $20.00 | 77 | 98.87s |
| 11 | GPT-6 Astra (max) OpenAI | 88 | 93 | $10.00 | $50.00 | — | — |
| 12 | GPT-6 Astra (xhigh) OpenAI | 88 | 93 | $10.00 | $50.00 | — | — |
| 13 | 88 | 92 | $10.00 | $50.00 | 55 | 8.71s | |
| 14 | Grok 4.6 (xhigh) SpaceXAI | 87 | 91 | $2.00 | $6.00 | 59 | 37.66s |
| 15 | GPT-6 Astra (high) OpenAI | 87 | 92 | $10.00 | $50.00 | — | — |
Models are scored using a weighted combination of benchmarks, pricing, and speed metrics relevant to this use case.
Models specifically designed for reasoning (like OpenAI o-series and DeepSeek R1) typically score highest on benchmarks like GPQA, AIME, and HLE. Check the rankings above for the latest results.
For tasks requiring genuine multi-step logic — math proofs, complex analysis, scientific research — yes. For simpler tasks, general-purpose models are more cost-effective.
Chain-of-thought (CoT) is when a model shows its step-by-step thinking process. Some models do this internally (hidden tokens), while others expose it. CoT generally improves accuracy on complex problems but increases token usage.