Which LLM is smartest, and which one is worth what it costs? This chart plots 29 of the 30 large language models we track by their intelligence score and their price per million tokens, so you can compare AI models from 14 labs at a glance. Look top-left for the best LLM for the price, top-right for the frontier, and click any model to try it against others on your own prompt.
Hover a dot for details. Click it to start a comparison with that model.
Intelligence: the Artificial Analysis Intelligence Index at each model's highest effort. Models without a score yet sit in the lane under the chart, placed by price. Price: blended per 1M tokens, 3 parts input to 1 part output. Lines at $2 blended and the median score (39.5).
The chart splits at $2 per 1M tokens and at the median intelligence score of 39.5. Where a model lands tells you what it is good for.
Score 39.5 or higher and cost under $2 per 1M tokens. The best value when you need quality on a budget.
Score 39.5 or higher and cost $2 or more. The most capable models, for hard reasoning, coding and agents.
Score under 39.5 and cost under $2. Cheap and fast for high-volume work like extraction, tagging and chat.
Score under 39.5 and cost $2 or more. Something cheaper scores as well or better, unless you need a specific feature.
Not scored yet: Gemma 4 31B IT. They sit in the "No score yet" lane under the chart, placed by price, and in the table below.
Every model with its intelligence score, blended price, context window and release month, ranked by score. Click a column to sort by it, or pick a model to open it in the side-by-side comparison.
| Claude Opus 5.5 | Anthropic | 57.6 | $8 | 1M | Sep 2026 | Closed |
|---|---|---|---|---|---|---|
| Claude Sonnet 5.5 | Anthropic | 56 | $4 | 1M | Sep 2026 | Closed |
| Claude Fable 5.1 | Anthropic | 53.4 | $20 | 1M | Sep 2026 | Closed |
| GPT-6 Astra | Open AI | 52.7 | $20 | 1.05M | Sep 2026 | Closed |
| Gemini 4 Argon | 52.6 | $4 | 1.05M | Oct 2026 | Closed | |
| GPT-6.1 Sol | Open AI | 51.8 | $4 | 1.05M | Sep 2026 | Closed |
| Grok 4.7 | XAI | 46.4 | $3 | 500K | Sep 2026 | Closed |
| MiMo V2.6 Pro | Xiaomi | 46.3 | $0.54 | 1.05M | Sep 2026 | Open |
| GLM 5.3 | Z.ai | 44.8 | $2.15 | 1M | Aug 2026 | Open |
| Kimi K3 | Kimi | 43.6 | $6 | 1.05M | Jul 2026 | Open |
| Claude Haiku 5.5 | Anthropic | 43.4 | $0.20 | 1M | Oct 2026 | Closed |
| GLM 5.3 Flash | Z.ai | 41.8 | $0.24 | 1M | Aug 2026 | Open |
| Gemini 3.8 Flash | 40.9 | $1.50 | 1.05M | Sep 2026 | Closed | |
| Qwen3.8 2.4T A95B | Alibaba | 39.9 | $3 | 262K | Aug 2026 | Open |
| DeepSeek V4.1 Flash | DeepSeek | 39.5 | $0.26 | 1M | Sep 2026 | Open |
| Muse Spark Open | Meta | 38.9 | $2 | 1.05M | Oct 2026 | Open |
| Mistral Large 4 | Mistral AI | 38.4 | $1.03 | 1.05M | Oct 2026 | Closed |
| GPT-6 Luna | Open AI | 38.1 | $0.20 | 1.05M | Sep 2026 | Closed |
| MiMo V2.6 Flash | Xiaomi | 37.9 | $0.18 | 1.05M | Sep 2026 | Open |
| DeepSeek V4 Pro | DeepSeek | 36 | $0.99 | 1M | Aug 2026 | Open |
| Qwen3.8 27B | Alibaba | 33.7 | $1.13 | 1M | Aug 2026 | Open |
| MiniMax M3 | MiniMax | 29.2 | $0.52 | 512K | Jun 2026 | Open |
| Kimi K2.7 Code | Kimi | 25.8 | $1.71 | 262K | Jun 2026 | Open |
| Inkling Small | Thinking Machines | 25.7 | $0.64 | 1M | Jul 2026 | Open |
| Nemotron 3 Ultra | NVIDIA | 22.9 | $1.05 | 1M | Jun 2026 | Open |
| Gemini 3.5 Flash Lite | 22.2 | $0.85 | 1.05M | Jul 2026 | Closed | |
| Mistral Medium 3.5 | Mistral AI | 14.2 | $3 | 262K | Apr 2026 | Open |
| GPT OSS 120B | Open AI | 11.6 | $0.20 | 131K | Aug 2025 | Open |
| Mistral Small 4 | Mistral AI | 11.3 | $0.26 | 262K | Mar 2026 | Open |
| Gemma 4 31B IT | — | $0.21 | 262K | Apr 2026 | Open |
Of the 30 models we track, Claude Opus 5.5 from Anthropic has the highest Artificial Analysis Intelligence Index score at 57.6, for $8 per 1M tokens blended. Next are Claude Sonnet 5.5 (56) and Claude Fable 5.1 (53.4).
Counting only models that score at or above the median of 39.5, Claude Haiku 5.5 is the cheapest at $0.20 per 1M tokens blended, with a score of 43.4.
Under $2 per 1M tokens, MiMo V2.6 Pro scores highest at 46.3, for $0.54 per 1M tokens blended. That is 80% of Claude Opus 5.5's score at 7% of its price.
MiMo V2.6 Pro from Xiaomi is the highest-scoring open-weights model, at 46.3, for $0.54 per 1M tokens blended.
Prices are blended per 1M tokens, weighted 3 parts input to 1 part output: (3 × input price + output price) ÷ 4. For example, Claude Opus 5.5 charges $4 per 1M input tokens and $20 per 1M output tokens, which blends to $8. The chart splits models at $2 blended and at the median intelligence score.