Model cost comparison.

Pounds per million tokens, cheapest blended first. Blended assumes a 3:1 input-heavy workload: (3 × input + output) ÷ 4.

Prices change often — last updated 2026-07-18, converted at $1 = £0.79. No live scraping; the table is an editable file.

ModelProviderTier£ in / 1M£ out / 1MBlended £ / 1M
mistral-smallMIMistralsmall£0.08£0.24£0.12
gpt-4.1-nanoOAOpenAIsmall£0.08£0.32£0.14
gemini-2.5-flashGOGooglesmall£0.12£0.47£0.21
llama-4-maverickOROpenRoutermid£0.16£0.47£0.24
codestralMIMistralsmall£0.24£0.71£0.36
deepseek-v3DSDeepSeeksmall£0.21£0.87£0.38
llama-3.3-70bOROpenRoutermid£0.47£0.62£0.51
gpt-4.1-miniOAOpenAImid£0.32£1.26£0.55
deepseek-r1DSDeepSeekmid£0.43£1.73£0.76
kimi-k2OROpenRoutermid£0.43£1.90£0.80
claude-haiku-3.5ANAnthropicsmall£0.63£3.16£1.26
o4-miniOAOpenAImid£0.87£3.48£1.52
qwen-2.5-maxOROpenRoutermid£1.26£5.06£2.21
mistral-largeMIMistralmid£1.58£4.74£2.37
gemini-2.5-proGOGooglefrontier£0.99£7.90£2.72
gpt-4.1OAOpenAIfrontier£1.58£6.32£2.77
gpt-4oOAOpenAIfrontier£1.98£7.90£3.46
claude-sonnet-4ANAnthropicmid£2.37£11.85£4.74
grok-3XAxAIfrontier£2.37£11.85£4.74
claude-opus-4ANAnthropicfrontier£11.85£59.25£23.70

Prices are the menu. Your bill is the restaurant. TokenLedger prices what you actually used →