1. Compare
  2. OpenAI vs Claude vs Gemini APIs for a chat feature

OpenAI vs Claude vs Gemini APIs for a chat feature

We priced each provider's small model and mid-tier model for the same chat. OpenAI's and Anthropic's small models cost the same per token. Google's mid-tier model is the cheapest of the three until its price rises in January. We compare price only, not how good the answers are.

Priced from vendor pages
Prices checked
20 Oct 2026
Exchange rate
1 USD = £0.7489
Sources
4
How we check

The answer

For 10,000 chats a month on a small model, OpenAI and Claude cost £3.37 and Gemini £9.36. On mid-tier models, OpenAI and Claude cost £67.40 and Gemini £25.28, rising to £50.55 from 1 January 2027.

What does an AI chat feature cost each month in API fees?

OpenAI

Small: gpt-6-luna ($0.10 in / $0.50 out per million tokens). Mid: gpt-6.1-sol ($2 / $10)

  • The lowest small-model price of the three current-generation models compared here, tied with Claude, and cached input from a tenth of the input price (a twentieth on gpt-6.1-sol).

Prices are for short prompts. Long-context requests on the flagship models cost more.

1,000 chats a month, small model £0.3410,000 chats a month, small model £3.3710,000 chats a month, mid-tier model £67.40

Claude (Anthropic)

Small: Haiku 5.5 ($0.10 in / $0.50 out per million tokens, prompts up to 100K). Mid: Sonnet 5.5 ($2 / $10)

  • The lowest small-model price of the three current-generation models compared here, tied with OpenAI, and half-price batch processing for work that can wait.

Haiku 5.5 costs five times more for prompts over 100,000 tokens ($0.50 in / $2.50 out).

1,000 chats a month, small model £0.3410,000 chats a month, small model £3.3710,000 chats a month, mid-tier model £67.40

Gemini (Google)

Small: gemini-3.1-flash-lite ($0.25 in / $1.50 out per million tokens). Mid: gemini-3.8-flash ($0.75 / $3.75, then $1.50 / $7.50 from 1 Jan 2027)

  • The cheapest mid-tier model for now, and a free tier on both models for testing.

Free-tier data is used to improve Google's products; paid data isn't. Gemini 3.8 Flash doubles in price on 1 January 2027.

1,000 chats a month, small model £0.9410,000 chats a month, small model £9.3610,000 chats a month, mid-tier model £25.28
1,000 chats a month, small model£0.34
1,000 chats a month, small model£0.34
1,000 chats a month, small model£0.94
10,000 chats a month, small model£3.37
10,000 chats a month, small model£3.37
10,000 chats a month, small model£9.36
10,000 chats a month, mid-tier model£67.40
10,000 chats a month, mid-tier model£67.40
10,000 chats a month, mid-tier model£25.28

Monthly cost in pounds, from each vendor's pricing page. Lowest figures compared price only, at our token assumption.

How we worked it out

  • Our assumption for one chat: 2,000 input tokens (the conversation so far plus your instructions) and 500 output tokens. Real chats vary, and long conversations send more input each turn.
  • OpenAI gpt-6-luna and Claude Haiku 5.5: 2,000 × $0.10 + 500 × $0.50 per million = $0.00045 a chat. 1,000 chats = $0.45 × 0.7489 = £0.34; 10,000 = $4.50 × 0.7489 = £3.37.
  • Gemini 3.1 Flash-Lite: 2,000 × $0.25 + 500 × $1.50 per million = $0.00125 a chat. 1,000 chats = $1.25 × 0.7489 = £0.94; 10,000 = $12.50 × 0.7489 = £9.36.
  • OpenAI gpt-6.1-sol and Claude Sonnet 5.5: 2,000 × $2 + 500 × $10 per million = $0.009 a chat. 10,000 = $90 × 0.7489 = £67.40.
  • Gemini 3.8 Flash: 2,000 × $0.75 + 500 × $3.75 per million = $0.003375 a chat. 10,000 = $33.75 × 0.7489 = £25.28. From 1 Jan 2027: 2,000 × $1.50 + 500 × $7.50 = $0.00675 a chat, so $67.50 × 0.7489 = £50.55.
  • Which models we compare: each provider's newest small and mid-tier model on standard paid rates. Google still sells the older gemini-2.5-flash-lite at $0.10 in / $0.40 out per million tokens, which is cheaper than all three small models here: 2,000 × $0.10 + 500 × $0.40 per million = $0.0004 a chat, so 1,000 chats = $0.40 × 0.7489 = £0.30 and 10,000 = $4.00 × 0.7489 = £3.00 (Gemini API pricing, https://ai.google.dev/gemini-api/docs/pricing, checked 23 Oct 2026).
  • Small and mid-tier are our grouping by price, not the providers' labels. We don't rank answer quality.
  • Standard rates only. Excludes VAT, prompt caching discounts, batch discounts and Google Search grounding.

Sources