AI Prediction Markets: Model Prices and Usage
Markets on where AI model API prices and usage go next, resolved from named public data. SF·SIX runs them on the Hedera testnet with test funds.
Testnet only. SF·SIX prediction markets run on the Hedera testnet with test funds that have no real value. No real money can be won or lost, and nothing here is investment advice.
283 AI models currently meet the rule for model-price ladders: at least one paid, standard, unconditional USD price on OpenRouter. Each market asks whether a model’s average provider price (or its daily OpenRouter token usage) will be at or above a threshold on a close day, and resolves from that day’s archived source data.
Models with price ladders
| Model | Average input price / 1M | Average output price / 1M | Qualifying endpoints | OpenRouter tokens (2026-10-16) | Markets |
|---|---|---|---|---|---|
| DeepSeek V4.1 FlashDeepSeek | $0.2304 | $0.9506 | 25 in / 25 out | 5.41T | Open markets |
| GLM 5.3 FlashZ.ai (Zhipu) | $0.1307 | $0.5583 | 25 in / 25 out | 1.73T | Open markets |
| MiMo-V2.6-FlashXiaomi | $0.1392 | $0.2933 | 6 in / 6 out | 1.44T | Open markets |
| Hy4 previewTencent | $0.834 | $2.50 | 3 in / 3 out | 1.24T | Open markets |
| GLM 5.3Z.ai (Zhipu) | $0.9879 | $4.16 | 33 in / 33 out | 942.62B | Open markets |
| Claude Opus 5.5Anthropic | $4.20 | $21.00 | 10 in / 10 out | 933.37B | Open markets |
| DeepSeek V4 Flash 0731DeepSeek | $0.1464 | $0.5528 | 19 in / 19 out | 620.6B | Open markets |
| Hy3Tencent | $0.1575 | $0.65 | 4 in / 4 out | 477.68B | Open markets |
| Muse Spark 1.3 ContributorMeta | $0.10 | $0.20 | 1 in / 1 out | 459.17B | Open markets |
| DeepSeek V4 Flash 0423DeepSeek | $0.1877 | $0.5241 | 11 in / 11 out | 423.23B | Open markets |
| Claude Sonnet 5.5Anthropic | $2.07 | $10.38 | 8 in / 8 out | 396.5B | Open markets |
| Qwen3.6 35B A3BQwen (Alibaba) | $0.154 | $1.11 | 10 in / 10 out | 386.14B | Open markets |
| Nemotron 3.5 Super VLNVIDIA | $0.15 | $0.61 | 5 in / 5 out | 287.58B | Open markets |
| MiniMax M3MiniMax | $0.3464 | $1.40 | 11 in / 11 out | 210.13B | Open markets |
| MiMo-V2.6-ProXiaomi | $0.4338 | $0.87 | 4 in / 4 out | 209.34B | Open markets |
| Kimi K3Moonshot AI | $1.83 | $14.32 | 19 in / 19 out | 203.41B | Open markets |
| Step 5 PreviewStepFun | $1.00 | $2.70 | 4 in / 4 out | 179.04B | Open markets |
| DeepSeek V4 Pro 0813DeepSeek | $0.9797 | $3.66 | 15 in / 15 out | 161.08B | Open markets |
| GLM 5.2Z.ai (Zhipu) | $0.9756 | $4.59 | 22 in / 22 out | 140.87B | Open markets |
| Nemotron 3 SuperNVIDIA | $0.155 | $1.08 | 3 in / 3 out | 105.18B | Open markets |
| Gemini 2.5 Flash LiteGoogle DeepMind | $0.10 | $0.40 | 3 in / 3 out | 102.54B | Open markets |
| Gemini 3 Flash PreviewGoogle DeepMind | $0.50 | $3.00 | 2 in / 2 out | 97.9B | Open markets |
| Nemotron 3.5 LightningNVIDIA | $0.0517 | $0.16 | 3 in / 3 out | 90.07B | Open markets |
| Claude Sonnet 5Anthropic | $2.10 | $10.50 | 10 in / 10 out | 87.16B | Open markets |
| gpt-oss-120bOpenAI | $0.1038 | $0.4599 | 23 in / 23 out | 86.23B | Open markets |
| DeepSeek V4 Pro 0423DeepSeek | $1.35 | $3.04 | 12 in / 12 out | 85.51B | Open markets |
| Nemotron 3 UltraNVIDIA | $0.5813 | $2.53 | 4 in / 4 out | 83.11B | Open markets |
| Claude Opus 5Anthropic | $5.25 | $26.25 | 10 in / 10 out | 82.77B | Open markets |
| Qwen3.8 FlashQwen (Alibaba) | $0.15 | $0.47 | 2 in / 2 out | 81.75B | Open markets |
| MiMo-V2.5Xiaomi | $0.40 | $2.00 | 1 in / 1 out | 79.12B | Open markets |
| Gemma 4 31BGoogle DeepMind | $0.2718 | $0.5655 | 11 in / 11 out | 73.13B | Open markets |
| Gemini 3.1 Flash LiteGoogle DeepMind | $0.2625 | $1.58 | 4 in / 4 out | 73.07B | Open markets |
| Qwen3.8 27BQwen (Alibaba) | $0.2452 | $2.30 | 16 in / 16 out | 62.19B | Open markets |
| Aion 3.5AionLabs | $3.00 | $6.00 | 1 in / 1 out | Not in named rows | Open markets |
| Aion 3.5 MiniAionLabs | $0.70 | $1.40 | 1 in / 1 out | Not in named rows | Open markets |
| Aion-2.0AionLabs | $0.80 | $1.60 | 1 in / 1 out | Not in named rows | Open markets |
| Aion-3.0AionLabs | $3.00 | $6.00 | 1 in / 1 out | Not in named rows | Open markets |
| Aion-3.0-MiniAionLabs | $0.70 | $1.40 | 1 in / 1 out | Not in named rows | Open markets |
| Aion-RP 1.0 (8B)AionLabs | $0.80 | $1.60 | 1 in / 1 out | Not in named rows | Open markets |
| Beam 1.0 501B A23BReflection | $0.7375 | $2.38 | 4 in / 4 out | Not in named rows | Open markets |
| Claude Fable 5Anthropic | $10.17 | $50.83 | 6 in / 6 out | Not in named rows | Open markets |
| Claude Fable 5.1Anthropic | $10.00 | $50.00 | 4 in / 4 out | Not in named rows | Open markets |
| Claude Haiku 4.5Anthropic | $1.05 | $5.25 | 8 in / 8 out | Not in named rows | Open markets |
| Claude Opus 4.1Anthropic | $15.00 | $75.00 | 1 in / 1 out | Not in named rows | Open markets |
| Claude Opus 4.5Anthropic | $5.08 | $25.42 | 6 in / 6 out | Not in named rows | Open markets |
| Claude Opus 4.6Anthropic | $5.08 | $25.42 | 6 in / 6 out | Not in named rows | Open markets |
| Claude Opus 4.7Anthropic | $5.19 | $25.94 | 8 in / 8 out | Not in named rows | Open markets |
| Claude Opus 4.8Anthropic | $5.25 | $26.25 | 10 in / 10 out | Not in named rows | Open markets |
| Claude Sonnet 4.5Anthropic | $3.00 | $15.00 | 1 in / 1 out | Not in named rows | Open markets |
| Claude Sonnet 4.6Anthropic | $3.13 | $15.67 | 9 in / 9 out | Not in named rows | Open markets |
The current value of each price metric, computed from the live catalog with the published rule; the settled value uses the archived capture at close. Most-used first; the 50 most-used of 283 are listed, the rest are in the model markets app. Usage ladders also need the model in OpenRouter’s named daily rows.
How the markets work
Each model has strike ladders: a set of yes/no questions at different thresholds for its average input price, average output price and daily token usage, each closing on a fixed UTC day. Before you trade, every market shows its exact question, data source, observation window, cutoff and versioned resolution rule. A proposed result is not final: challenges, or missing or conflicting evidence, can delay or void resolution.
The usage numbers come from the same OpenRouter rankings as the most-used AI models ranking; the price numbers from the same endpoint prices as the AI API pricing tables.
Frequently asked questions
What are AI prediction markets on SF·SIX?
Yes/no markets on measurable facts about AI models: whether a model’s average provider price or its daily token usage will be at or above a threshold on a given day. They run on the Hedera testnet with test funds that have no real value.
How do model price markets resolve?
On the close day, the market takes the simple mean of the standard USD price per 1M tokens of every OpenRouter provider endpoint that publishes a paid, numeric, unconditional price for the model, from the last archived capture of that day. Free, variable, conditional or tiered, non-USD and unavailable endpoints are excluded, and no qualifying provider voids the market.
How do model usage markets resolve?
On OpenRouter’s published daily token total for the model, counting its routing variants with it. If the model is not in OpenRouter’s named daily rows, the usage market is void rather than resolved as zero.
Can I win real money?
No. SF·SIX prediction markets run on the Hedera testnet with test funds that have no real value. Nothing on this page is investment advice.
Compute markets
SF·SIX does not list GPU or compute-hour markets and holds no GPU rental price data. Its markets cover the inference layer: per-token API prices and token usage.