Guides / Pricing

What 1,000 AI API requests cost on 19 models (October 2026)

Updated November 5, 2026. Researched and drafted with AI help; every figure is checked against the linked source.

Per-million-token prices are hard to picture. This page turns them into something concrete: what 1,000 requests cost on every current model we track, for three common kinds of work. Every number comes from the provider’s official pricing page (linked below) and is calculated the same way for every model.

The three workloads

• Chat: 2,000 input tokens (system prompt, history, question) and 500 output tokens. A typical chatbot or support reply.
• Coding agent step: 20,000 input tokens, of which 18,000 are a repeated context read from the prompt cache, plus 1,000 output tokens. One step of an agent that re-reads the same codebase each turn.
• Summary: 10,000 input tokens (a long document) and 300 output tokens.

The formula is simply tokens × price per million, times 1,000 requests. For the agent column, the 18,000 cached tokens are charged at the model’s cache read price.

Results

Sorted by the chat column, cheapest first. Prices are per million tokens (input / output).

* Mistral Large 3’s cache read price of $0.05 per million comes from Mistral’s API pricing docs; earlier versions of this table charged its cached tokens at the full input price ($11.50 per 1,000 agent steps). Grok 4.7’s cache read price of $0.50 per million comes from xAI’s separate pricing page; earlier versions of this table charged it at the full input price ($46.00 per 1,000 agent steps) because xAI’s models page lists no cache price.

† Gemini 3.8 Flash prices are promotional through December 31, 2026. From January 1, 2027 Google lists $1.50 input, $7.50 output and $0.15 cache reads, which doubles its row.

‡ GPT-5.6 Sol is shown at the $5 / $30 that OpenAI’s main pricing page listed until it stopped showing per-token prices. OpenAI’s developer pricing page and model page show a promotional $4 / $20 that OpenAI says is available until November 30, 2026. At $4 / $20 (and $0.40 cache reads) its row would be $18.00, $35.20 and $46.00.

§ Claude Haiku 5.5 (released October 7, 2026) has two price tiers: $0.10 / $0.50 for prompts up to 100,000 tokens, and $0.50 / $2.50 above that. All three workloads here stay under 100,000 tokens. It succeeds Claude Haiku 4.5 ($1 / $5), whose rows would be $4.50, $8.80 and $11.50.

° Gemini 4 Argon (generally available since October 13, 2026) is at an introductory price through November 30, 2026, with cache reads at $0.10. From December 1, Google lists $4 / $20 and $0.20 cache reads, which doubles its row to $18.00, $31.60 and $46.00, the same as Claude Opus 5.5. Argon supports only implicit caching, so its agent figure assumes the repeated context is served from cache. See Gemini 4 Argon explained.

¶ Mistral Large 4 is in public preview. When Mistral released its open weights in late October 2026 it cut the API price to $0.70 / $2.10 with $0.07 cache reads, according to its announcement and API pricing docs. At the earlier $1.36 / $4.18 ($0.14 cache reads) its rows were $4.81, $9.42 and $14.85, so the cut roughly halves them and moves it from 10th to 7th cheapest for chat.

♦ GPT-6.1 Terra was released on October 26, 2026 and succeeds GPT-5.6 Terra ($2 / $12), which OpenAI’s pricing page no longer lists and whose rows were $10.00, $19.60 and $23.60. Like other OpenAI models, it charges 2x input and 1.5x output for prompts over 272K tokens.

What stands out

The spread is about 100x. The same 1,000 chats cost $0.45 on GPT-6 Luna or Claude Haiku 5.5 and $45 on GPT-6 Astra or Claude Fable 5.1. For classification, extraction or simple replies, a budget model is usually the right starting point.

Output price drives the chat column. Even though a chat sends four times more tokens than it receives, output is priced about 3 to 8 times higher, so it is often half the bill or more. Capping reply length is one of the easiest savings.

Caching reorders the agent column. In an agent step most input is cached, so the cache read price matters more than the headline input price. Claude Opus 5.5 lists input at $4 but cache reads at $0.20 per million (95% off), which keeps its agent step at $31.60. Among the flagships, GPT-6 Astra’s cache read of $1 per million makes it the most expensive agent step at $88, compared with $74.50 for Claude Fable 5.1 at the same headline price. The $2 / $10 models now tie: Anthropic cut Claude Sonnet 5.5’s cache read from $0.20 to $0.10 per million on October 7, 2026, matching GPT-6.1 Sol, so both cost $15.80 per 1,000 agent steps (Sonnet 5.5 was $17.60 before the cut). Gemini 4 Argon joins them at the same $15.80 while its introductory price lasts. Grok 4.7’s cache read is a smaller discount (75% off, $0.50 per million), so it slips from 11th cheapest for chat to 14th for agent steps, at $19.00. GPT-6.1 Terra’s $0.05 cache read (95% off) keeps it at $7.90 per 1,000 agent steps, about half of what the $2 / $10 models cost.

Same price, different token count. Anthropic says Claude 4.7 and later models use a newer tokenizer that produces about 30% more tokens for the same text than earlier Claude models. So Claude Sonnet 5.5 and GPT-6.1 Sol, both $2 / $10, may not cost the same for the same prompt, and neither may Claude Haiku 5.5 and GPT-6 Luna at $0.10 / $0.50. Test with your own text before deciding.

Ways to pay less than the table

• Batch. Anthropic, OpenAI and Google offer a Batch API at 50% off if you can wait for results, and Mistral lists the same 50% batch discount. For those models, every figure above would halve. xAI’s pricing page now lists a smaller 20% batch discount for Grok 4.7, which would cut its row to $5.60, $15.20 and $17.44. We have not found a published batch discount for DeepSeek.
• Off-peak DeepSeek. DeepSeek’s prices above are peak prices. Peak hours are 01:00–04:00 and 06:00–10:00 UTC on weekdays; at all other times, including weekends and Chinese public holidays, it charges half. See AI API pricing explained for the details.
• Long context costs more on some models. Gemini 3.1 Pro and Grok 4.7 double their input price above 200K tokens, and Claude Haiku 5.5 charges five times more above 100K tokens. None of the workloads here come close, but a large agent context can.

Run your own numbers

These three workloads are only examples. The cost calculator lets you enter your own token counts, cache share and batch setting for any model, and gives you a link you can share. For a head-to-head view, use the model comparisons, and if a model you depend on seems off today, check the Dumb meter to see whether other people are noticing the same thing.

Prices change often. We recheck each provider’s official pricing page regularly and update this page when they change.

Sources

1. Anthropic: Claude API pricing
2. OpenAI API pricing
3. OpenAI developer docs: pricing (GPT-6 models)
4. Google: Gemini API pricing
5. xAI: Models and pricing
6. xAI docs: Pricing
7. DeepSeek API pricing
8. Mistral AI pricing
9. Mistral docs: Mistral Large 4
10. Google: Gemini 4 Argon is now available for everyone
11. GPT-6.1 Terra model page
12. Mistral: Mistral Large 4 open weights announcement
13. Mistral docs: API pricing
14. Anthropic: Claude Haiku 5.5