← All Tools

Best Claude Haiku 4.5 Alternatives

Claude Haiku 4.5 by Anthropic is a budget model priced at $0.8/4 per 1M tokens (in/out). Looking for a better deal or different capabilities? Here are the best options.

Claude Haiku 4.5

AnthropicBudget

Input

$0.8/1M

Output

$4/1M

Context

200K

Max Output

8K

Why Switch from Claude Haiku 4.5?

Weakest reasoning in the Claude family
Can struggle with nuanced instructions

Top Alternatives

#1Gemini 3 FlashGoogleBudget

27% cheaper, higher benchmark scores, 1M context (5x more).

Input

$0.5/1M

38% cheaper

Output

$3/1M

25% cheaper

Context

1M

Max Output

66K

MMLU-Pro: 78%(+8.6%)HumanEval: 90%(+1.9%)
#2Mistral Medium 3MistralMid-Tier

50% cheaper, higher benchmark scores.

Input

$0.4/1M

50% cheaper

Output

$2/1M

50% cheaper

Context

128K

Max Output

16K

MMLU-Pro: 76%(+6.6%)HumanEval: 87%(-1.1%)
#3Gemini 2.5 FlashGoogleBudget

Dramatically cheaper (84% less), higher benchmark scores, 1M context (5x more).

Input

$0.15/1M

81% cheaper

Output

$0.6/1M

85% cheaper

Context

1M

Max Output

66K

MMLU-Pro: 76%(+6.6%)HumanEval: 89.5%(+1.4%)
#4GPT-4o MiniOpenAIBudget

Dramatically cheaper (84% less), comparable performance.

Input

$0.15/1M

81% cheaper

Output

$0.6/1M

85% cheaper

Context

128K

Max Output

16K

MMLU-Pro: 68%(-1.4%)HumanEval: 87.2%(-0.9%)
#5o3OpenAIReasoning

Dramatically cheaper (58% less), higher benchmark scores, 100K max output.

Input

$0.4/1M

50% cheaper

Output

$1.6/1M

60% cheaper

Context

200K

Max Output

100K

MMLU-Pro: 87%(+17.6%)HumanEval: 94.5%(+6.4%)
#6GLM-4.7Zhipu AIMid-Tier

42% cheaper, higher benchmark scores, 128K max output.

Input

$0.6/1M

25% cheaper

Output

$2.2/1M

45% cheaper

Context

200K

Max Output

128K

MMLU-Pro: 84.3%(+14.9%)HumanEval:
#7GLM-5Zhipu AIFlagship

12% cheaper, comparable performance, 128K max output.

Input

$1/1M

25% more

Output

$3.2/1M

20% cheaper

Context

200K

Max Output

128K

MMLU-Pro: 70.4%(+1.0%)HumanEval: 91%(+2.9%)
#8o4-miniOpenAIReasoning

Higher benchmark scores, 100K max output, adds reasoning.

Input

$1.1/1M

38% more

Output

$4.4/1M

10% more

Context

200K

Max Output

100K

MMLU-Pro: 85%(+15.6%)HumanEval: 93.5%(+5.4%)

Full Comparison Table

ModelInput $/1MOutput $/1MContextMMLU-ProHumanEvalScore
Gemini 3 FlashGoogle$0.5038% cheaper$3.0025% cheaper1M78%+8.6%90%+1.9%94
Mistral Medium 3Mistral$0.4050% cheaper$2.0050% cheaper128K76%+6.6%87%-1.1%85
Gemini 2.5 FlashGoogle$0.1581% cheaper$0.6085% cheaper1M76%+6.6%89.5%+1.4%84
GPT-4o MiniOpenAI$0.1581% cheaper$0.6085% cheaper128K68%-1.4%87.2%-0.9%83
o3OpenAI$0.4050% cheaper$1.6060% cheaper200K87%+17.6%94.5%+6.4%79
GLM-4.7Zhipu AI$0.6025% cheaper$2.2045% cheaper200K84.3%+14.9%73
GLM-5Zhipu AI$1.0025% more$3.2020% cheaper200K70.4%+1.0%91%+2.9%72
o4-miniOpenAI$1.1038% more$4.4010% more200K85%+15.6%93.5%+5.4%69
Mistral Large 3Mistral$2.00150% more$5.0025% more128K83%+13.6%91%+2.9%69
Llama 4 MaverickMeta$0.3161% cheaper$0.8579% cheaper1M80.5%+11.1%90.2%+2.1%68
Llama 4 ScoutMeta$0.1878% cheaper$0.6384% cheaper10M74.2%+4.8%86%-2.1%68
DeepSeek R1DeepSeek$0.5531% cheaper$2.1945% cheaper128K84%+14.6%92%+3.9%67
GPT-5OpenAI$1.2556% more$10.00150% more128K88.5%+19.1%95%+6.9%65
Gemini 3.1 ProGoogle$2.00150% more$12.00200% more1M91%+21.6%95%+6.9%65
Gemini 3 ProGoogle$2.00150% more$12.00200% more1M89.8%+20.4%94%+5.9%65
Gemini 2.5 ProGoogle$1.2556% more$10.00150% more1M87.5%+18.1%93.5%+5.4%65
GPT-4oOpenAI$2.50213% more$10.00150% more128K80.5%+11.1%91%+2.9%62
Claude Opus 4.6Anthropic$5.00525% more$25.00525% more200K89.5%+20.1%95%+6.9%59
Claude Sonnet 4.6Anthropic$3.00275% more$15.00275% more200K86%+16.6%94%+5.9%59
Claude Sonnet 4.5Anthropic$3.00275% more$15.00275% more200K84.5%+15.1%93%+4.9%59
GPT-5.3 CodexOpenAI$2.00150% more$16.00300% more200K90%+20.6%96.5%+8.4%59
GPT-5.2 CodexOpenAI$1.75119% more$14.00250% more200K89%+19.6%95.5%+7.4%59
MiniMax M2.5MiniMax$0.3063% cheaper$1.2070% cheaper200K82%+12.6%90%+1.9%57
Grok 4xAI$3.00275% more$15.00275% more128K86%+16.6%93%+4.9%55
DeepSeek V3DeepSeek$0.1483% cheaper$0.2893% cheaper164K78%+8.6%89%+0.9%47

Head-to-Head Comparisons

Alternatives for Other Models