API pricing
API rates are shown per million tokens. Cached-input rates, peak and off-peak windows, context thresholds, and deployment regions remain separate; prices stay in their listed currencies without conversion. Missing values are labeled “No comparable data.”
| Provider / model | Pricing tier / region | Input | Cached input | Output | Conditions and source |
|---|---|---|---|---|---|
OpenAI · GPT-6 Astragpt-6-astra | Standard price | $10 | No comparable data | $50 | Per 1M tokens; the official model directory does not list a cached input price in this summary row. Official API pricing · Checked 2026-10-21 |
OpenAI · GPT-6.1 Solgpt-6.1-sol | Standard price | $2 | No comparable data | $10 | Per 1M tokens; a cached input price is not listed in this official model-directory summary row. Official API pricing · Checked 2026-10-21 |
Anthropic · Claude Opus 5.5claude-opus-5-5 | Standard price | $4 | No comparable data | $20 | Per 1M tokens; see the official full price list for caching, batch, and service-tier prices. Official API pricing · Checked 2026-10-21 |
Anthropic · Claude Fable 5.1claude-fable-5-1 | Standard price | $10 | No comparable data | $50 | Per 1M tokens; see the official full price list for caching, batch, and service-tier prices. Official API pricing · Checked 2026-10-21 |
Anthropic · Claude Sonnet 5.5claude-sonnet-5-5 | Standard price | $2 | No comparable data | $10 | Per 1M tokens; see the official full price list for caching, batch, and service-tier prices. Official API pricing · Checked 2026-10-21 |
Anthropic · Claude Haiku 5.5claude-haiku-5-5 | Starting standard price | $0.1 | No comparable data | $0.5 | Per 1M tokens; the official model overview marks this as a starting price. See the full price list for cache pricing. Official API pricing · Checked 2026-10-21 |
Google · Gemini 4 Argongemini-4-argon | Introductory price (through 2026-11-30) | $2 | $0.1 | $10 | Per 1M tokens; from 2026-12-01, input is $4, cached input $0.20, and output $20. Explicit context caching is not supported. Official API pricing · Checked 2026-10-21 |
DeepSeek · DeepSeek-V4.1 Flashdeepseek-flash | Peak hours | $0.3 | $0.006 | $1.2 | Per 1M tokens; cache-miss input is $0.30. Weekdays 01:00–04:00 and 06:00–10:00 UTC, excluding Chinese public holidays. Official API pricing · Checked 2026-10-21 |
DeepSeek · DeepSeek-V4.1 Flashdeepseek-flash | Off-peak hours | $0.15 | $0.003 | $0.6 | Per 1M tokens; the provider says off-peak prices are half the peak prices. Official API pricing · Checked 2026-10-21 |
Alibaba Cloud Qwen · Qwen3.8 Maxqwen3.8-max | Real-time API calls | ¥12 | No comparable data | ¥36 | Per 1M tokens; context caching discounts are available, but the price table does not list the cache-hit rate. Official API pricing · Checked 2026-10-21 |
xAI · Grok 4.7grok-4.7 | Short context (under 200K) | $2 | $0.5 | $6 | Per 1M tokens; the US regional endpoint adds 10% to the standard price. Official API pricing · Checked 2026-10-21 |
xAI · Grok 4.7grok-4.7 | Long context (200K and above) | $4 | $1 | $12 | Per 1M tokens; once a prompt reaches the threshold, all tokens in that request use long-context prices. Official API pricing · Checked 2026-10-21 |
Notes before comparing
Section titled “Notes before comparing”- DeepSeek lists separate peak and off-peak rates; use the UTC hours shown on its official page.
- Qwen rates vary by deployment region. The official page may list an international-region rate in CNY.
- xAI applies its long-context rate to the entire request once input reaches 200K tokens; US-region endpoints have a separate surcharge.
- Google Gemini 4 Argon’s introductory rate has a stated end date; its standard rate takes effect on 2026-12-01.
- API token prices are usage charges. They are not the same as allowances included with chat or team subscriptions.