Gemini 4 Argon
Claude Opus 5.5
Muse Wonder
MiMo-V2.6-Pro
Qwen3.8-Max
Kimi K3
GLM-5.3
GPT-5.5
ERNIE 5.1
Grok 4.20 Beta
amazon-nova-experimental-chat
Dola Seed 2.0 Pro
NVIDIA-Nemotron-3-Ultra-550B-A55B
Gemma 4 31B
Step 5 Preview
DeepSeek-V3
DeepSeek-V3 is an open-weight model from DeepSeek, released Mar 24, 2025. It ranks #181 of 351 on the LMArena text leaderboard with a rating of 1375. It costs $0.29 per million input tokens and $1.14 per million output tokens, with a 160K-token context window.
- Developer
- DeepSeek
- Released
- Mar 24, 2025
- Context window
- 160K
- Price in / out, per 1M tokens
- $0.29 / $1.14
- Weights
- Open · DeepSeek
- Inputs
- Text
- Reasoning
- No
- Tool use
- Yes
- Hugging Face downloads, 30 days
- 659.6K
- Hugging Face likes
- 3.2K
The rating LMArena published at each update since Dec 30, 2024. Ratings move as votes come in and new models join.
Benchmarks
| Benchmark | Score bar | Accuracy | Rank | Cost / test |
|---|---|---|---|---|
| LiveCodeBenchCoding | 65.48% | 101 / 143 | $0.9 / $0.9 | |
| MedQAMedicine | 80.90% | 69 / 92 | $0.9 / $0.9 | |
| GPQA DiamondReasoning | 54.55% | 118 / 138 | $0.9 / $0.9 | |
| MMLU-ProReasoning | 73.82% | 116 / 138 | $0.9 / $0.9 | |
| AIMEReasoning | 27.50% | 74 / 96 | $0.9 / $0.9 |
Vals AI, as of Oct 24, 2026.
LMArena, by category
| Leaderboard | Rating | Rank | Votes |
|---|---|---|---|
| TextLMArena | 1375 | 181 / 351 | 45K |
| CodingLMArena | 1368 | 209 / 346 | 8.3K |
| MathLMArena | 1373 | 194 / 339 | 3.2K |
| Creative writingLMArena | 1364 | 151 / 349 | 6K |
| Instruction followingLMArena | 1345 | 194 / 351 | 12.4K |
| Hard promptsLMArena | 1365 | 194 / 351 | 18.6K |
| Multi-turnLMArena | 1389 | 168 / 349 | 7.7K |
| Long queriesLMArena | 1352 | 200 / 330 | 8.5K |
| ExpertLMArena | 1351 | 201 / 305 | 2.4K |
| Legal and governmentIndustries | 1379 | 181 / 325 | 2.9K |
| Medicine and healthcareIndustries | 1383 | 178 / 325 | 2.7K |
| Business and financeIndustries | 1349 | 195 / 344 | 6.9K |
| ChineseLanguages | 1390 | 188 / 328 | 3K |
| SpanishLanguages | 1358 | 186 / 249 | 987 |
| GermanLanguages | 1374 | 149 / 261 | 1.2K |
| FrenchLanguages | 1385 | 168 / 252 | 594 |
| JapaneseLanguages | 1333 | 140 / 236 | 1.1K |
LMArena, CC BY 4.0, as of Oct 22, 2026. Where a model is rated at several effort levels, its best is shown (DeepSeek-V3 (0324), DeepSeek-V3).
Compare DeepSeek-V3
DeepSeek
A research lab known for strong open-weight reasoning and coding models trained at low cost. Founded 2023, based in Hangzhou. deepseek.com
Questions
Who makes DeepSeek-V3?
DeepSeek-V3 is made by DeepSeek, founded 2023 and based in Hangzhou.
How much does DeepSeek-V3 cost?
$0.29 per million input tokens and $1.14 per million output tokens, at its provider’s list price.
What is DeepSeek-V3’s context window?
160K tokens (163,840).
How does DeepSeek-V3 rank?
#181 of 351 on LMArena’s text leaderboard, rated 1375 (95% interval 1371–1379) from 44,974 votes, as of Oct 22, 2026.
Is DeepSeek-V3 open source?
DeepSeek-V3 has open weights under the DeepSeek license, downloadable from Hugging Face (deepseek-ai/DeepSeek-V3-0324).