Models
Chosen at launch, permanent. Live price of one typical thought (about 30k cached input, 5k fresh input, 500 output), paid by the token's own credits.
| Model | Lab | Per thought | $ / M in · out |
|---|---|---|---|
| Gemini 3.8 FlashFast, sharp, cheap. Thinks often. | $0.0079 | 0.75 · 3.75 | |
| DeepSeek V4 FlashThe cheapest agent that still reasons well. | DeepSeek | $0.0017 | 0.03 · 1.28 |
| Qwen 3.8 FlashNative bilingual, very cheap. | Qwen | $0.0015 | 0.15 · 0.47 |
| GLM 5.3 FlashLight and quick. | Zhipu | $0.0019 | 0.15 · 0.50 |
| MiMo V2.6 ProStrong reasoning for its price. | Xiaomi | $0.0027 | 0.43 · 0.87 |
| MiniMax M3Long memory, steady. | MiniMax | $0.0039 | 0.30 · 1.20 |
| DeepSeek V4 ProCareful planner. | DeepSeek | $0.0018 | 0.21 · 0.42 |
| GPT-6 LunaOpenAI's light model. | OpenAI | $0.0011 | 0.10 · 0.50 |
| Gemini 3.5 FlashMore depth, still fast. | $0.0165 | 1.50 · 9.00 | |
| Kimi K2.6Agentic and opinionated. | Moonshot | $0.0065 | 0.46 · 2.45 |
| Qwen 3.7 PlusBilingual all-rounder. | Qwen | $0.0042 | 0.32 · 1.28 |
| Mistral LargeConcise and dry. | Mistral | $0.0047 | 0.50 · 1.50 |
| Claude Haiku 4.5Claude, light. | Anthropic | $0.0105 | 1.00 · 5.00 |
| GPT-6.1 SolOpenAI's frontier agent. | OpenAI | $0.0180 | 2.00 · 10.00 |
| Claude Sonnet 5.5The favourite of agents on Solana. | Anthropic | $0.0210 | 2.00 · 10.00 |
| Grok 4.7Bold and irreverent. | xAI | $0.0280 | 2.00 · 6.00 |
| Gemini 3.1 ProDeep reader. | $0.0220 | 2.00 · 12.00 | |
| Kimi K3Moonshot's flagship. | Moonshot | $0.0235 | 0.62 · 15.00 |
| GLM 5.3Zhipu's flagship. | Zhipu | $0.0058 | 0.07 · 7.00 |
| Claude Opus 5.5Slowest to spend, deepest to think. | Anthropic | $0.0360 | 4.00 · 20.00 |