Difficult prompts that separate frontier models from mid-tier ones. Ranked by Arena text / hard_prompts, priced against live API rates. 44 models have a published score on this board.
Capability published 2026-09-02 · pricing retrieved 2026-09-10
anthropic/claude-fable-5.1google/gemini-3.8-flashdeepseek/deepseek-v4-flash-0731google/gemini-3.8-flashRows marked frontier are Pareto-optimal — nothing we track is both stronger and cheaper. "Strength" rescales the board so 100% = parity with the leader.
| # | Model | Provider | Score | Strength | $/1M | Context | Tools |
|---|---|---|---|---|---|---|---|
| 1 | Claude Fable 5.1anthropic/claude-fable-5.1 | Anthropic | 1528 | 100% | $20.00 | 1M | tools |
| 2 | Claude Opus 5anthropic/claude-opus-5 | Anthropic | 1526 | 100% | $10.00 | 1M | tools |
| 3 | Claude Opus 4.6anthropic/claude-opus-4.6 | Anthropic | 1523 | 99% | $10.00 | 1M | tools |
| 4 | Claude Fable 5anthropic/claude-fable-5 | Anthropic | 1511 | 95% | $20.00 | 1M | tools |
| 5 | Gemini 3.8 Flashgoogle/gemini-3.8-flash | 1508 | 94% | $1.50 | 1.048576M | tools | |
| 6 | Claude Opus 4.7anthropic/claude-opus-4.7 | Anthropic | 1499 | 92% | $10.00 | 1M | tools |
| 7 | Kimi K3moonshotai/kimi-k3 | Moonshot AI | 1497 | 91% | $6.00 | 1.048576M | tools |
| 8 | Gemini 3.7 Flashgoogle/gemini-3.7-flash | 1497 | 91% | $1.50 | 1.048576M | tools | |
| 9 | Qwen3.8 Max (0902)qwen/qwen3.8-max-0902 | Qwen (Alibaba) | 1494 | 90% | $3.00 | 1M | tools |
| 10 | GLM 5.3z-ai/glm-5.3 | Z.ai (Zhipu) | 1491 | 90% | $2.15 | 1.31072M | tools |
| 11 | Gemini 3.5 Flashgoogle/gemini-3.5-flash | 1490 | 89% | $3.38 | 1.048576M | tools | |
| 12 | GLM 5.3 Flashz-ai/glm-5.3-flash | Z.ai (Zhipu) | 1486 | 88% | $0.24 | 1.31072M | tools |
| 13 | Claude Sonnet 4.6anthropic/claude-sonnet-4.6 | Anthropic | 1485 | 88% | $6.00 | 1M | tools |
| 14 | Gemini 3.6 Flashgoogle/gemini-3.6-flash | 1484 | 88% | $1.50 | 1.048576M | tools | |
| 15 | Qwen3.7 Maxqwen/qwen3.7-max | Qwen (Alibaba) | 1482 | 87% | $2.21 | 1M | tools |
| 16 | GPT-5.6 Solopenai/gpt-5.6-sol | OpenAI | 1478 | 86% | $4.00 | 1.05M | tools |
| 17 | GLM 5.2z-ai/glm-5.2 | Z.ai (Zhipu) | 1476 | 85% | $1.49 | 1.048576M | tools |
| 18 | Claude Opus 4.8anthropic/claude-opus-4.8 | Anthropic | 1472 | 84% | $10.00 | 1M | tools |
| 19 | GLM 5.1z-ai/glm-5.1 | Z.ai (Zhipu) | 1472 | 84% | $1.49 | 205K | tools |
| 20 | GPT-5.6 Terraopenai/gpt-5.6-terra | OpenAI | 1470 | 83% | $4.50 | 1.05M | tools |
| 21 | Kimi K2.6moonshotai/kimi-k2.6 | Moonshot AI | 1469 | 83% | $1.71 | 262K | tools |
| 22 | Grok 4.5x-ai/grok-4.5 | xAI | 1468 | 83% | $3.00 | 500K | tools |
| 23 | Claude Sonnet 5anthropic/claude-sonnet-5 | Anthropic | 1462 | 81% | $4.00 | 1M | tools |
| 24 | DeepSeek V4 Pro 0813deepseek/deepseek-v4-pro-0813 | DeepSeek | 1461 | 81% | $1.57 | 1.048576M | tools |
| 25 | DeepSeek V4 Pro 0423deepseek/deepseek-v4-pro | DeepSeek | 1461 | 81% | $1.20 | 1.048576M | tools |
Capability: Arena leaderboard dataset (CC-BY-4.0), Arena text / hard_prompts, published 2026-09-02. Pricing: OpenRouter, retrieved 2026-09-10. (input x 3 + output x 1) / 4.
curl -s https://agentleaderboards.com/api/v1/route/reasoning.json | jq .picksNo key, open CORS, refreshed daily. Set your own budget and constraints →
Writing & editing code · Autonomous agent work · Tool-call reliability · Following operator instructions · Recovering from shell errors · Mathematics · Precise instruction following · Long multi-turn chat · Long inputs · Creative writing · General assistant · Software & IT services · Legal & government · Medicine & healthcare · Science · Business & finance · Writing & language