Alternatives
Best cheaper alternatives to Claude Haiku
Looking for a cheaper or smarter replacement for Claude Haiku by Anthropic? We compared every LLM model with verifiable pricing on real input/output cost, context window, intelligence and modalities. 18 options are meaningfully cheaper.
Category: LLM
Llama 3.1 8B Instant 128k97% CHEAPER
Groq
Same category
$0.05 in · $0.08 out / 1M Tokens 128K ctx 98%
Gemini 1.5 Flash-8B97% CHEAPER
Google
Same category
$0.0375 in · $0.15 out / 1M Tokens 36 1.0M ctx 98%
Ministral 3 3B95% CHEAPER
Mistral AI
Same category
$0.1 in · $0.1 out / 1M Tokens 128K ctx 98%
Qwen3.5 9B94% CHEAPER
Together AI
Same category
$0.1 in · $0.15 out / 1M Tokens 262K ctx 98%
OpenAI GPT OSS 20B94% CHEAPER
Fireworks AI
Same category
$0.07 in · $0.3 out / 1M Tokens 128K ctx 98%
Gemini 1.5 Flash93% CHEAPER
Google
Same category
$0.075 in · $0.3 out / 1M Tokens 48 1.0M ctx 98%
GPT OSS 20B 128k93% CHEAPER
Groq
Same category
$0.075 in · $0.3 out / 1M Tokens 128K ctx 98%
Ministral 3 8B93% CHEAPER
Mistral AI
Same category
$0.15 in · $0.15 out / 1M Tokens 37 128K ctx 98%
Llama 4 Scout 17Bx16E 128k92% CHEAPER
Groq
Same category
$0.11 in · $0.34 out / 1M Tokens 128K ctx 98%
DeepSeek V4 Flash91% CHEAPER
DeepSeek
Same category
$0.14 in · $0.28 out / 1M Tokens 128K ctx 98%
Ministral 3 14B90% CHEAPER
Mistral AI
Same category
$0.2 in · $0.2 out / 1M Tokens 128K ctx 98%
Mistral Small93% CHEAPER
Mistral AI
Same category
$0.1 in · $0.3 out / 1M Tokens 44 128K ctx 95%
Llama 3.1 8B97% CHEAPER
Meta
Same category
$0.055 in · $0.055 out / 1M Tokens 34 128K ctx 85%
Llama 3.2 11B Vision97% CHEAPER
Meta
Same category
$0.055 in · $0.055 out / 1M Tokens 38 128K ctx 85%
Imagen 498% CHEAPER
Google
Same category
$0.04 in · $0.04 out / image 80%
DeepSeek-V3 (Chat)94% CHEAPER
DeepSeek
Same category
$0.07 in · $0.27 out / 1M Tokens 63 64K ctx 80%
Gemini 2.0 Flash91% CHEAPER
Google
Same category
$0.1 in · $0.4 out / 1M Tokens 56 1.0M ctx 80%
Qwen Image Plus99% CHEAPER
Alibaba Cloud
Same category
$0.03 in · $0.03 out / image 78%
Costs are normalised to a blended 1M-token workload (or per-unit price for non-token models) using each model's latest verified pricing. Models with pricing under review are excluded from recommendations.