AI model API prices, compared
Per-1M-token list prices for 25 models from 8 providers. Every price links to the provider's own pricing page. Last checked 2026-10-14.
25 models · 8 providers · 25 of 25 prices verified against source · 12 head-to-head comparisons · Cheapest input today: $0.075/1M (GPT-OSS 20B on Groq)
Counts come from our public prices.json. Last checked 2026-10-14.
Prices sourced from the official pricing pages of: Anthropic · DeepSeek · Google · Groq · Mistral · OpenAI · Together · xAI
We aren't affiliated with or endorsed by these providers.
Last price check: 2026-10-14 · Price changes recorded so far: 21 · Follow changes via RSS (changelog)
Use the data directly, no key needed:
curl -s https://aipricedrops.com/prices.jsonAlso available: prices.csv · llms.txt for AI agents
Every price change is logged with its date, old price and new price. See the changelog.
| Provider | Model | Input $/M ⇅ | Output $/M ⇅ | Context ⇅ | Last verified |
|---|---|---|---|---|---|
| OpenAI | GPT-6 Astra | $10.0 | $50.0 | — | 2026-10-14 |
| OpenAI | GPT-6.1 Sol | $2.0 | $10.0 | — | 2026-10-14 |
| OpenAI | GPT-6 Luna | $0.1 | $0.5 | — | 2026-10-14 |
| OpenAI | GPT-5.6 Sol | $4.0 | $20.0 | — | 2026-10-14 |
| Anthropic | Claude Fable 5.1 | $10.0 | $50.0 | — | 2026-10-14 |
| Anthropic | Claude Opus 5.5 | $4.0 | $20.0 | — | 2026-10-14 |
| Anthropic | Claude Sonnet 5.5 | $2.0 | $10.0 | — | 2026-10-14 |
| Anthropic | Claude Haiku 5.5 | $0.5 | $2.5 | — | 2026-10-14 |
| xAI | Grok 4.7 | $2.0 | $6.0 | 500000 | 2026-10-14 |
| xAI | Grok 4.3 | $1.25 | $2.5 | 1000000 | 2026-10-14 |
| xAI | Grok Build 0.1 | $1.0 | $2.0 | 256000 | 2026-10-14 |
| Gemini 3.1 Pro Preview | $2.0 | $12.0 | — | 2026-10-14 | |
| Gemini 3.8 Flash | $0.75 | $3.75 | — | 2026-10-14 | |
| Gemini 3.5 Flash-Lite | $0.3 | $2.5 | — | 2026-10-14 | |
| Gemini 3.1 Flash-Lite | $0.25 | $1.5 | — | 2026-10-14 | |
| DeepSeek | DeepSeek V4.1 Flash | $0.3 | $1.2 | 1000000 | 2026-10-14 |
| DeepSeek | DeepSeek V4 Pro | $1.32 | $3.96 | 1000000 | 2026-10-14 |
| Mistral | Mistral Large 4 | $0.68 | $2.09 | 1000000 | 2026-10-14 |
| Mistral | Mistral Large 3 | $0.5 | $1.5 | — | 2026-10-14 |
| Mistral | Mistral Small 4 | $0.15 | $0.6 | — | 2026-10-14 |
| Groq | GPT-OSS 120B (Groq) | $0.15 | $0.6 | 131072 | 2026-10-14 |
| Groq | GPT-OSS 20B (Groq) | $0.075 | $0.3 | 131072 | 2026-10-14 |
| Together | Qwen3.8 Flash (Together) | $0.15 | $0.47 | — | 2026-10-14 |
| Together | GLM-5.3 Flash (Together) | $0.15 | $0.5 | — | 2026-10-14 |
| Together | Kimi K3 (Together) | $3.0 | $15.0 | — | 2026-10-14 |
Cheapest for this job
“Tier” is our rough capability judgment (1 = light, 4 = frontier), not a benchmark. Picks are the cheapest models at or above the tier.
Simple Q&A / chat
Needs tier ≥1; example 10M in / 2M out per month- GPT-OSS 20B (Groq) — $1.35/mo
- GPT-6 Luna — $2.00/mo
- Qwen3.8 Flash (Together) — $2.44/mo
Caveat: Small models are fine for FAQs and short answers but hallucinate more on niche facts.
Full ranking for simple q&a / chat →Coding
Needs tier ≥2; example 20M in / 5M out per month- GLM-5.3 Flash (Together) — $5.50/mo
- GPT-OSS 120B (Groq) — $6.00/mo
- DeepSeek V4.1 Flash — $12.00/mo
Caveat: Coding quality varies a lot by language and repo size; run your own tests before switching.
Full ranking for coding →Long documents
Needs tier ≥2; example 100M in / 2M out per month- DeepSeek V4.1 Flash — $32.40/mo
- Mistral Large 4 — $72.18/mo
- Grok 4.3 — $130.00/mo
Caveat: Only models with a published context ≥500K are considered; several providers bill more above 200K–272K prompt tokens.
Full ranking for long documents →Hard reasoning
Needs tier ≥3; example 10M in / 10M out per month- Mistral Large 4 — $27.70/mo
- DeepSeek V4 Pro — $52.80/mo
- Grok 4.7 — $80.00/mo
Caveat: Reasoning models spend extra 'thinking' tokens billed as output, so real cost can be several times this estimate.
Full ranking for hard reasoning →Savings calculator
FAQ
- How do AI model API prices work?
- Providers bill separately for input tokens (your prompt) and output tokens (the model reply), usually in US dollars per 1 million tokens. TokenTally shows the standard pay-as-you-go list rate: no batch discount, no cached input, short-context prompts.
- How often does TokenTally update prices?
- We re-check official provider pricing pages and rebuild the table regularly. This snapshot was last verified on 2026-10-14. Every change is recorded on the price-change log and in the RSS feed at /changes.xml.
- Which AI model API is cheapest right now?
- Among the 25 models we track, the lowest input list price is $0.075/1M for GPT-OSS 20B (Groq) from Groq. For a workload match, use the Cheapest for this job pages or the savings calculator — the cheapest overall model is not always the best fit.
- Where do the prices come from?
- Every row links to the provider's official pricing page (the src URL in /prices.json). We mark a row verified only after reading that page. Always check the official pricing page before you buy.
- Can AI agents download the price table?
- Yes. Prefer /prices.json or /prices.csv over scraping HTML, and start with /llms.txt for an agent-oriented map of the site. Prices may be wrong or out of date — check the provider src URL before relying on them.