Live LLM pricing - every model, every provider, one table
Input, output and cached token prices across providers, verified against primary sources. No sign-up, no client-side data grids.
Data as of 2026-08-01 · 338 models tracked
338
models
50
providers
18
MCP passports
7 🔴
traps found
Top price movements
Price history tracking since 2026-07
We record every change from the moment we start watching a source. No confirmed price changes yet - this is where they will appear.
What connecting an MCP server really costs
Keys, free-tier limits, traps and the schema token tax - cost passports for popular MCP servers, every number with a proof and a date.
Open the passports18
passports
7 🔴
traps found
6
run for $0
Cheapest models by input price
| Model | Context | Input$ / 1M | Output$ / 1M | Cached$ / 1M |
|---|---|---|---|---|
| Ling-2.6-flashinclusionAI | 262K | 0.01 | 0.03 | 0.002 |
| Granite 4.0 MicroIBM | 131K | 0.017 | 0.112 | - |
| Mistral NemoMistral | 131K | 0.02 | 0.04 | - |
| Nex-N2-MiniNex AGI | 262K | 0.025 | 0.1 | 0.0025 |
| Llama 3.2 1B InstructMeta | 131K | 0.027 | 0.201 | - |
| gpt-oss-20bOpenAI | 131K | 0.03 | 0.13 | 0.03 |
| Nova Micro 1.0Amazon | 128K | 0.035 | 0.14 | - |
| gpt-oss-120bOpenAI | 131K | 0.037 | 0.17 | - |
What would it cost you?
Pick a task and a monthly volume - we show the three cheapest models.
- Ling-2.6-flash$3.40 per month
- Mistral Nemo$5.60 per month
- Llama 3 8B Lunaris$9.40 per month
Estimated for 200000 requests/month at 800 input / 300 output tokens per request. Open full calculator →