LLM API pricing: official, OpenRouter and KovAPI
The same chat models at the vendor's own list price, at OpenRouter and on KovAPI, per 1M input and output tokens. OpenRouter charges the vendor's list price for most of these models; KovAPI is below it on every row, and every price links to the page it came from.
Input, per 1M tokens
| Tier | Official | OpenRouter | KovAPI |
|---|---|---|---|
| Claude Opus 5.5 | $4.00 ↗ | $4.00 ↗ | $2.50save 37% vs official$2.2727 with the 10% bonus on a $1,250 top-up, 43% below official |
| Claude Sonnet 5.5 | $2.00 ↗ | $2.00 ↗ | $1.50save 25% vs official$1.3636 with the 10% bonus on a $1,250 top-up, 31% below official |
| GPT-6.2 Astra | $8.00 ↗ | $8.00 ↗ | $2.40save 70% vs official$2.1818 with the 10% bonus on a $1,250 top-up, 72% below official |
| GPT-6.2 Sol | $2.00 ↗ | $2.00 ↗ | $0.60save 70% vs official$0.5455 with the 10% bonus on a $1,250 top-up, 72% below official |
| GPT-6 Astra | $10.00 ↗ | $10.00 ↗ | $3.00save 70% vs official$2.7273 with the 10% bonus on a $1,250 top-up, 72% below official |
| Gemini 4 Hydrogen | $1.50 ↗ | $1.50 ↗ | $0.45save 70% vs official$0.4091 with the 10% bonus on a $1,250 top-up, 72% below official |
| Gemini 3.8 Flash | $0.75 ↗until 2026-12-31; $1.50 / $7.50 from 2027-01-01 | $0.825 ↗ | $0.60save 20% vs official$0.5455 with the 10% bonus on a $1,250 top-up, 27% below official |
| Grok 4.7 | $2.00 ↗prompts under 200K tokens; $4.00 / $12.00 at 200K and above | $2.00 ↗ | $0.30save 85% vs official$0.2727 with the 10% bonus on a $1,250 top-up, 86% below official |
| DeepSeek V4.1 Flash | $0.15 ↗off-peak rate; peak hours (01:00–04:00 and 06:00–10:00 UTC, Mon–Fri) $0.30 / $1.20 | $0.15 ↗ | $0.075save 50% vs official$0.0682 with the 10% bonus on a $1,250 top-up, 54% below official |
per 1M tokens
Output, per 1M tokens
| Tier | Official | OpenRouter | KovAPI |
|---|---|---|---|
| Claude Opus 5.5 | $20.00 ↗ | $20.00 ↗ | $12.50save 37% vs official$11.3636 with the 10% bonus on a $1,250 top-up, 43% below official |
| Claude Sonnet 5.5 | $10.00 ↗ | $10.00 ↗ | $7.50save 25% vs official$6.8182 with the 10% bonus on a $1,250 top-up, 31% below official |
| GPT-6.2 Astra | $40.00 ↗ | $40.00 ↗ | $12.00save 70% vs official$10.9091 with the 10% bonus on a $1,250 top-up, 72% below official |
| GPT-6.2 Sol | $10.00 ↗ | $10.00 ↗ | $3.00save 70% vs official$2.7273 with the 10% bonus on a $1,250 top-up, 72% below official |
| GPT-6 Astra | $50.00 ↗ | $50.00 ↗ | $15.00save 70% vs official$13.6364 with the 10% bonus on a $1,250 top-up, 72% below official |
| Gemini 4 Hydrogen | $7.50 ↗ | $7.50 ↗ | $2.25save 70% vs official$2.0455 with the 10% bonus on a $1,250 top-up, 72% below official |
| Gemini 3.8 Flash | $3.75 ↗until 2026-12-31; $1.50 / $7.50 from 2027-01-01 | $4.125 ↗ | $3.60save 4% vs official$3.2727 with the 10% bonus on a $1,250 top-up, 12% below official |
| Grok 4.7 | $6.00 ↗prompts under 200K tokens; $4.00 / $12.00 at 200K and above | $6.00 ↗ | $0.90save 85% vs official$0.8182 with the 10% bonus on a $1,250 top-up, 86% below official |
| DeepSeek V4.1 Flash | $0.60 ↗off-peak rate; peak hours (01:00–04:00 and 06:00–10:00 UTC, Mon–Fri) $0.30 / $1.20 | $0.60 ↗ | $0.30save 50% vs official$0.2727 with the 10% bonus on a $1,250 top-up, 54% below official |
per 1M tokens
KovAPI measured on real calls
Counted: real customer requests. Not counted: our own tests, requests rejected for client errors (4xx) and content the model refused. Failed calls are never billed. Updated every 10 minutes.
Notes
- Standard tier, prompts of normal length. Vendors charge more for long context (Grok 4.7: $4.00 / $12.00 per 1M at 200K prompt tokens and above) and less for batch jobs (OpenAI and Google: half price); those tiers are not in the table.
- DeepSeek's official price has two rates: the table shows the off-peak rate; at peak hours (01:00–04:00 and 06:00–10:00 UTC, Monday to Friday) it is $0.30 / $1.20.
- Gemini 3.8 Flash: Google's price is $0.75 / $3.75 until 2026-12-31 and $1.50 / $7.50 from 2027-01-01.
- All models here work with the same OpenAI-compatible KovAPI key; Claude models also speak the native Anthropic Messages API.
How we compare
- List prices only: the price a new account pays today, without volume deals or one-off coupons. A promotion with an end date is shown with that date.
- Video is per second of finished video at 16:9 with no reference video; prices quoted per clip are divided by the clip length and the original price is shown under it. Images are per picture at each resolution.
- Token-billed models (the official ByteDance price, OpenRouter) are converted with the vendor's own formula: tokens per second = width x height x 24 / 1024.
- KovAPI prices are read live from our billing table, the same numbers a request is charged. Competitor and official prices are rechecked at least once a month, and each one shows the date it was last checked.
- “With the top-up bonus” = list price ÷ (1 + bonus), at the largest bonus: 10% on a $1,250 top-up. Bonuses start at $50 (3%); smaller top-ups get none. Bonus credits are spent like paid credits.
FAQ
Does OpenRouter charge the official price?
How much does the Claude Opus 5.5 API cost?
What does the GPT-6.2 API cost?
Why is DeepSeek shown at the off-peak price?
How often are these prices checked?
One OpenAI-compatible key for every model on this page. Pay as you go; failed jobs are refunded in full.
Get an API key Claude Opus 5.5Claude Sonnet 5.5