Claude Fable 5.1 vs. the closest competing models - Claude Fable 5.1 vs Claude Fable 5 vs GPT-5.6 Sol vs Kimi K3 vs Claude Opus 5

Pricing

Metric Type	Claude Fable 5.1Anthropic · via OpenRouter	Claude Fable 5Anthropic · via OpenRouter	GPT-5.6 SolOpenAI · via OpenRouter	Kimi K3Moonshot AI · via OpenRouter	Claude Opus 5Anthropic · via OpenRouter
Inputper 1M tokens	$10.00TextImageFile	$10.00TextImageFile	$4.00FileImageText	$0.24TextImage	$5.00TextImageFile
Outputper 1M tokens	$50.00Text	$50.00Text	$20.00Text	$10.80Text	$25.00Text
Cached inputper 1M tokens	$0.25	$1.00	$0.40	$0.21	$0.50
Cache writeper 1M tokens	$12.50	$12.50	$5.00	—	$6.25
Cache write (1h)per 1M tokens	$20.00	$20.00	—	—	$10.00
Web searchper request	$0.010	$0.010	$0.010	—	$0.010

This pricing matrix highlights the API costs, prompt caching tiers, and modality support for flagship frontier models available via OpenRouter:

Main highlights:

Claude Fable 5.1 and Claude Fable 5 sit at the high-end frontier tier with $10.00 / 1M input and $50.00 / 1M output tokens.
Major Prompt Caching Improvement in Fable 5.1: While both Fable versions share base input/output rates, Claude Fable 5.1 slashes cached input costs by 4× (down to $0.25 / 1M tokens compared to $1.00 / 1M on Fable 5), making persistent long-context sessions significantly more cost-effective.
Mid-Tier Balance with Claude Opus 5 (2× cheaper than Fable) & GPT-5.6 Sol delivers aggressive pricing for a frontier-grade model at $2.00 input / $10.00 output with low cache write overhead ($2.50 / 1M tokens).
Anthropic models uniquely offer extended 1-hour cache write tiers ($10.00–$20.00 / 1M tokens) for apps requiring longer persistent prompt memory across requests.
Grounded web search is uniformly priced at $0.010 per request across OpenAI and Anthropic models