Tokenando Indices
TTPI: Token Price Index
What buyers actually paid per million tokens, per lab: the day’s spend on each lab’s models divided by the tokens bought, so caching discounts and the real input/output mix are included. Free tiers and self-hosting are excluded. Labs with thin coverage drop for the day; they are never estimated.
Latest settled day 2026-10-11. One number per lab: what buyers paid for that lab's average routed token, with caching discounts included.
Settled 10 October 2026
$0.322per million tokens, realized-1.14%
| Lab | $/MTok (realized) | Δ day | $/MTok (list) | Below list | Models | Volume coverage | Settled |
|---|---|---|---|---|---|---|---|
| Anthropic | $0.904 | -3.64% | — | — | 16 | 100.0% | 2026-10-10 (behind the latest settled day, 2026-10-11) |
| xAI | $0.834 | +1.92% | — | — | 7 | 99.9% | 2026-10-10 (behind the latest settled day, 2026-10-11) |
| Moonshot AI | $0.619 | -2.77% | — | — | 7 | 100.0% | 2026-10-10 (behind the latest settled day, 2026-10-11) |
| $0.445 | -1.26% | — | — | 26 | 99.6% | 2026-10-10 (behind the latest settled day, 2026-10-11) | |
| OpenAI | $0.322 | -1.14% | — | — | 63 | 98.6% | 2026-10-10 (behind the latest settled day, 2026-10-11) |
| Qwen | $0.215 | +0.15% | — | — | 42 | 92.4% | 2026-10-10 (behind the latest settled day, 2026-10-11) |
| Meta | $0.183 | +1.28% | — | — | 8 | 100.0% | 2026-10-10 (behind the latest settled day, 2026-10-11) |
| Z.ai (Zhipu) | $0.138 | -3.28% | — | — | 16 | 100.0% | 2026-10-10 (behind the latest settled day, 2026-10-11) |
| Mistral | $0.130 | +3.28% | — | — | 20 | 98.4% | 2026-10-10 (behind the latest settled day, 2026-10-11) |
| DeepSeek | $0.050 | +1.40% | — | — | 14 | 100.0% | 2026-10-10 (behind the latest settled day, 2026-10-11) |
| MiniMax | $0.029 | -2.39% | — | — | 8 | 100.0% | 2026-10-10 (behind the latest settled day, 2026-10-11) |
| Tencent | $0.012 | -7.06% | — | — | 7 | 100.0% | 2026-10-10 (behind the latest settled day, 2026-10-11) |
- Dropped on 2026-10-11: Anthropic,
insufficient_models - Dropped on 2026-10-11: DeepSeek,
insufficient_models - Dropped on 2026-10-11: Google,
insufficient_models - Dropped on 2026-10-11: Meta,
insufficient_models - Dropped on 2026-10-11: MiniMax,
insufficient_models - Dropped on 2026-10-11: Mistral,
insufficient_models - Dropped on 2026-10-11: Moonshot AI,
insufficient_models - Dropped on 2026-10-11: OpenAI,
insufficient_models - Dropped on 2026-10-11: Qwen,
insufficient_models - Dropped on 2026-10-11: Tencent,
insufficient_models - Dropped on 2026-10-11: xAI,
insufficient_models - Dropped on 2026-10-11: Z.ai (Zhipu),
insufficient_models
List-price series
The same models priced at posted list rates instead: input at 75%, output at 25%, no caching, weighted by volume. Published since 2026-08-15; the realized series starts 2026-09-21, when per-model spend data began. The gap between the two is what caching and input-heavy traffic take off the rate card.
Settled 11 October 2026
$2.20per million tokens, list price+1.04%
GET /api/v1/ttpi?lab=anthropic&from=2026-09-21, or add &series=list for list prices. Free, no key, CORS open. spend and volume: OpenRouter rankings snapshot, disclosed bias
Tokenando indices are informational reference data, not financial products, price quotes, or investment advice. Offered prices are labeled offered; sources, weights, and exclusion rules are fully disclosed in the methodology. Settled values are immutable; corrections ship as new methodology versions, never as silent edits.