← All Tools

Best DeepSeek V3 Alternatives

DeepSeek V3 by DeepSeek is a open source model priced at $0.14/0.28 per 1M tokens (in/out). It's already affordable, but you might want different strengths or features.

DeepSeek V3

DeepSeekOpen Source

Input

$0.14/1M

Output

$0.28/1M

Context

164K

Max Output

16K

Why Switch from DeepSeek V3?

No vision support
Smaller context than competitors
China-based — availability concerns

Top Alternatives

#1MiniMax M2.5MiniMaxOpen Source

Higher benchmark scores, 128K max output, open-source and self-hostable.

Input

$0.3/1M

114% more

Output

$1.2/1M

329% more

Context

200K

Max Output

128K

MMLU-Pro: 82%(+4.0%)HumanEval: 90%(+1.0%)
#2Llama 4 MaverickMetaOpen Source

Comparable performance, 1M context (6x more), adds vision.

Input

$0.31/1M

121% more

Output

$0.85/1M

204% more

Context

1M

Max Output

32K

MMLU-Pro: 80.5%(+2.5%)HumanEval: 90.2%(+1.2%)
#3Llama 4 ScoutMetaOpen Source

10M context (61x more), adds vision, open-source and self-hostable.

Input

$0.18/1M

29% more

Output

$0.63/1M

125% more

Context

10M

Max Output

32K

MMLU-Pro: 74.2%(-3.8%)HumanEval: 86%(-3.0%)
#4DeepSeek R1DeepSeekReasoning

Comparable performance, 64K max output.

Input

$0.55/1M

293% more

Output

$2.19/1M

682% more

Context

128K

Max Output

64K

MMLU-Pro: 84%(+6.0%)HumanEval: 92%(+3.0%)
#5GLM-4.7Zhipu AIMid-Tier

Comparable performance, 128K max output, adds vision.

Input

$0.6/1M

329% more

Output

$2.2/1M

686% more

Context

200K

Max Output

128K

MMLU-Pro: 84.3%(+6.3%)HumanEval:
#6Gemini 2.5 FlashGoogleBudget

Comparable performance, 1M context (6x more), 66K max output.

Input

$0.15/1M

7% more

Output

$0.6/1M

114% more

Context

1M

Max Output

66K

MMLU-Pro: 76%(-2.0%)HumanEval: 89.5%(+0.5%)
#7o3OpenAIReasoning

Higher benchmark scores, 100K max output, adds vision, tool-use.

Input

$0.4/1M

186% more

Output

$1.6/1M

471% more

Context

200K

Max Output

100K

MMLU-Pro: 87%(+9.0%)HumanEval: 94.5%(+5.5%)
#8Gemini 3 FlashGoogleBudget

Comparable performance, 1M context (6x more), 66K max output.

Input

$0.5/1M

257% more

Output

$3/1M

971% more

Context

1M

Max Output

66K

MMLU-Pro: 78%(Same)HumanEval: 90%(+1.0%)

Full Comparison Table

ModelInput $/1MOutput $/1MContextMMLU-ProHumanEvalScore
MiniMax M2.5MiniMax$0.30114% more$1.20329% more200K82%+4.0%90%+1.0%80
Llama 4 MaverickMeta$0.31121% more$0.85204% more1M80.5%+2.5%90.2%+1.2%75
Llama 4 ScoutMeta$0.1829% more$0.63125% more10M74.2%-3.8%86%-3.0%68
DeepSeek R1DeepSeek$0.55293% more$2.19682% more128K84%+6.0%92%+3.0%58
GLM-4.7Zhipu AI$0.60329% more$2.20686% more200K84.3%+6.3%58
Gemini 2.5 FlashGoogle$0.157% more$0.60114% more1M76%-2.0%89.5%+0.5%56
o3OpenAI$0.40186% more$1.60471% more200K87%+9.0%94.5%+5.5%53
Gemini 3 FlashGoogle$0.50257% more$3.00971% more1M78%Same90%+1.0%53
Claude Opus 4.6Anthropic$5.003471% more$25.008829% more200K89.5%+11.5%95%+6.0%43
GPT-5.3 CodexOpenAI$2.001329% more$16.005614% more200K90%+12.0%96.5%+7.5%43
GPT-5.2 CodexOpenAI$1.751150% more$14.004900% more200K89%+11.0%95.5%+6.5%43
o4-miniOpenAI$1.10686% more$4.401471% more200K85%+7.0%93.5%+4.5%43
Mistral Large 3Mistral$2.001329% more$5.001686% more128K83%+5.0%91%+2.0%43
GPT-4o MiniOpenAI$0.157% more$0.60114% more128K68%-10.0%87.2%-1.8%42
GPT-5OpenAI$1.25793% more$10.003471% more128K88.5%+10.5%95%+6.0%40
Gemini 3.1 ProGoogle$2.001329% more$12.004186% more1M91%+13.0%95%+6.0%40
Gemini 3 ProGoogle$2.001329% more$12.004186% more1M89.8%+11.8%94%+5.0%40
Gemini 2.5 ProGoogle$1.25793% more$10.003471% more1M87.5%+9.5%93.5%+4.5%40
Grok 4xAI$3.002043% more$15.005257% more128K86%+8.0%93%+4.0%40
Mistral Medium 3Mistral$0.40186% more$2.00614% more128K76%-2.0%87%-2.0%40
GLM-5Zhipu AI$1.00614% more$3.201043% more200K70.4%-7.6%91%+2.0%38
Claude Sonnet 4.6Anthropic$3.002043% more$15.005257% more200K86%+8.0%94%+5.0%36
Claude Sonnet 4.5Anthropic$3.002043% more$15.005257% more200K84.5%+6.5%93%+4.0%36
Claude Haiku 4.5Anthropic$0.80471% more$4.001329% more200K69.4%-8.6%88.1%-0.9%30
GPT-4oOpenAI$2.501686% more$10.003471% more128K80.5%+2.5%91%+2.0%28

Head-to-Head Comparisons

Alternatives for Other Models