DeepSeek V3
DeepSeek V3Rankingin non-thinking mode?
Alternatives?
CheaperClaude Haiku 5.5xhigh×2-37%$0.12Remove sampling parameters+5
BetterGPT-5.6 Lunalow+36%+43%$0.01OpenDeepSeek V4 Pro 0813max×2+529%$0.67Memory (4-bit): 411 GB → 990 GB+5CheaperGPT-6 Lunamax×2-37%$0.068Function calling only with none · Chat Completions+3
CheaperGLM 5.3 FlashmaxOpen+92%-25%$0.28Defaults to max · Self-hosted+4Claude Haiku 5.5xhigh×2-37%$0.12Remove sampling parameters+5GPT-6 Lunamax×2-37%$0.068Function calling only with none · Chat Completions+3GLM 5.3 FlashmaxOpen+92%-25%$0.28Defaults to max · Self-hosted+4GPT OSS 120BhighOpen+82%-81%$0.092
Qwen3.7 Flashunstated+77%-83%$0.03 / $0.13GPT-5.6 Lunalow+36%+43%$0.01
Gemini 3.1 Flash Litehigh+64%+79%$0.06GPT-6 Lunamax×2-37%$0.068Function calling only with none · Chat Completions+3Claude Haiku 5.5high+99%-37%$0.079Remove sampling parameters+5GPT OSS 120BhighOpen+82%-81%$0.092DeepSeek V4 Pro 0813maxOpen×2+529%$0.67Memory (4-bit): 411 GB → 990 GB+5
Kimi K2.6unstatedOpen+97%+444%$0.95 / $4Memory (4-bit): 411 GB → 616 GBKimi K2.7 CodeunstatedOpen+95%+444%$0.95 / $4Memory (4-bit): 411 GB → 616 GB+1GLM 5.3 FlashmaxOpen+92%-25%$0.28Defaults to max · Self-hosted+4GLM 5.1unstatedOpen+91%+583%$1.40 / $4.40Memory (4-bit): 411 GB → 452 GBDeepSeek V4 Pro 0813maxOpen×2—$0.67Memory (4-bit): 411 GB → 990 GB+5R1 0528unstatedOpen+36%—$0.39 / $1.63DeepSeek R1 Distill Qwen 32BOpen+14%—$0.27 / $0.27No strict JSONDeepSeek R1 Distill Llama 70BOpen+5%—$0.70 / $0.80DeepSeek R1 0528 Qwen3 8BunstatedOpen-10%—$0.06 / $0.09Context: 128K → 33KDeepSeek V3 0324unstatedOpen-23%—$0.24 / $0.90DeepSeek V4.1 Flash 1017maxOpenn/a—$0.26Memory (4-bit): 411 GB → 458 GB+5DeepSeek OCR 2Openn/a—$0.03 / $0.03Context: 128K → 8KAll DeepSeek models →Run it yourself
estimateHistory
5 versionsBenchmarks
all categoriesFAQ
How much does DeepSeek V3 cost?›
$0.28 per 1M input tokens and $0.42 per 1M output tokens; cached input $0.028 per 1M. A typical benchmark task costs about $0.02 at default effort.
Is DeepSeek V3 free to use?›
There is no free API; the cheapest paid price is $0.28 per 1M input tokens. The weights are open, so you can run it yourself.
How good is DeepSeek V3, and what is it best at?›
#329 of 334 in Chat & Knowledge, #75 of 132 in Coding, #117 of 133 in Agents & Tool Use, #102 of 195 in Writing & Language, #122 of 133 in Vision & Documents.
Is DeepSeek V3 better than DeepSeek R1 Distill Llama 70B?›
On OTIS Mock AIME 2024 2025, DeepSeek V3 scores 48.9 vs 51.4 — lower.
How can I use DeepSeek V3?›
Through the API of DeepInfra, Novita.
What is the context window of DeepSeek V3?›
128K tokens, with up to 8K output tokens.
When was DeepSeek V3 released? Is there a newer version?›
Released 2024-12-26. It has been superseded by DeepSeek V3.1.
Is DeepSeek V3 open source?›
Yes, the weights are open.
Can I run DeepSeek V3 locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can DeepSeek V3 read images?›
No, it accepts text only.
What is DeepSeek V3?›
DeepSeek V3 is an open-weights language model from DeepSeek, released Dec 2024.
What are the best alternatives to DeepSeek V3?›
Claude Haiku 5.5 (cheaper), GPT-5.6 Luna (better), DeepSeek V4 Pro 0813 (open weights).