GPT, Claude, Gemini, Llama, Mistral, Qwen, DeepSeek and the rest. Grouped by the lab that builds them, with context window, modalities, licence and price laid out so you can pick the right model for the job.
The HokAI directory currently indexes 167 models. Every entry is independently verified against primary sources, re-verified on a rolling schedule, and graded across capability, pricing, deployment posture and compliance. Listing requires a flat submission fee, but ranking is governed by a public editorial methodology and cannot be purchased. Open a use case group below to see every listing in that area, or filter by a single facet such as free pricing, open source licence, SOC 2 attestation, HIPAA, EU AI Act or zero data retention.
84 records re-checked against vendor pages in the last 30 days
30 of 167 models
| Model | Creator | Blended $/M (3:1) | SWE-bench Verified | Frontier |
|---|---|---|---|---|
| Nemotron 3.5 Lightning | NVIDIA | $0.087/M | 51.56% | Yes |
| Nemotron 3 Nano 30B A3B | NVIDIA | $0.087/M | 38.76% | -- |
| Nemotron 3.5 Super VL 120B-A12B | NVIDIA | $0.175/M | 71.1% | Yes |
| Hy3 | Tencent | $0.282/M | 55% | -- |
| MAI-Code-1.1-Flash | Microsoft | $0.45/M | 72.6% | Yes |
| MiniMax M3 | MiniMax | $0.525/M | 75% | Yes |
| Solar Pro 4 | Upstage | $0.525/M | 70.6% | -- |
| DeepSeek-V4-Pro-0813 | DeepSeek | $0.544/M | 80.6% | Yes |
| Mistral Large 3 | Mistral AI | $0.75/M | 41.4% | -- |
| Seed 2.0 Pro | ByteDance | $0.945/M | 76.5% | -- |
| DeepSeek V4 | DeepSeek | $0.99/M | 80.6% | -- |
| Beam | Reflection AI | $1.50/M | 81.6% | Yes |
| Grok 4.20 | xAI | $1.56/M | 75% | -- |
| GLM-5.2 | Z.ai | $2.15/M | 82.8% | Yes |
| GPT-5.6 Luna | OpenAI | $2.25/M | 60.2% | -- |
| Grok 4.6 | xAI | $3.00/M | 95.6% | Yes |
| Gemini 3.5 Flash | $3.38/M | 78% | -- | |
| GPT-5.4 | OpenAI | $5.63/M | 80% | -- |
| GPT-5.6 Terra | OpenAI | $5.63/M | 72.3% | -- |
| Claude Sonnet 5 | Anthropic | $6.00/M | 82.1% | -- |
| Claude Sonnet 4.6 | Anthropic | $6.00/M | 79.6% | -- |
| Kimi K3 | Moonshot AI | $6.00/M | 67.5% | -- |
| Claude Opus 5 | Anthropic | $10.00/M | 97% | Yes |
| Claude Opus 4.8 | Anthropic | $10.00/M | 88.6% | -- |
| Claude Opus 4.7 | Anthropic | $10.00/M | 87.6% | -- |
| Claude Opus 4.6 | Anthropic | $10.00/M | 80.8% | -- |
| GPT-5.5 | OpenAI | $11.25/M | 88.7% | -- |
| GPT-5.6 Sol | OpenAI | $11.25/M | 78.5% | -- |
| Claude Fable 5 | Anthropic | $20.00/M | 95% | -- |
| GPT-5.5 Pro | OpenAI | $67.50/M | 88.7% | -- |
30 GA models carry both a vendor-checked price and a SWE-bench Verified score. Prices run from $0.087/M (Nemotron 3.5 Lightning) to $67.50/M (GPT-5.5 Pro); the highest score is 97% (Claude Opus 5) at $10.00/M.
Nine models are not beaten on both axes by any other: Nemotron 3.5 Lightning, Nemotron 3.5 Super VL 120B-A12B, MAI-Code-1.1-Flash, MiniMax M3, DeepSeek-V4-Pro-0813, Beam, GLM-5.2, Grok 4.6 and Claude Opus 5. Every other point is cheaper-or-better on at most one axis than something on that frontier.
Open the full model leaderboard