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-V4-Pro
DeepSeek-V4-Pro is an open-weight model from DeepSeek, released Aug 12, 2026. It ranks #49 of 351 on the LMArena text leaderboard with a rating of 1451, and #19 of 27 on the Vals Index with 52.97%. It costs $0.66 per million input tokens and $1.98 per million output tokens, with a 1M-token context window.
- Developer
- DeepSeek
- Released
- Aug 12, 2026
- Context window
- 1M
- Price in / out, per 1M tokens
- $0.66 / $1.98
- Weights
- Open · MIT
- Inputs
- Text
- Reasoning
- Yes
- Tool use
- Yes
- Hugging Face downloads, 30 days
- 151K
- Hugging Face likes
- 899
The rating LMArena published at each update since Apr 23, 2026. Ratings move as votes come in and new models join.
Benchmarks
| Benchmark | Score bar | Accuracy | Rank | Cost / test |
|---|---|---|---|---|
| Vals IndexVals AI | 52.97% | 19 / 27 | $2.50 | |
| Terminal-Bench 4.0Coding | 11.11% | 39 / 55 | $3.45 | |
| SWE-benchCoding | 77.40% | 35 / 85 | $0.44 | |
| Vibe Code BenchCoding | 49.93% | 58 / 115 | $2.21 | |
| Code MigrationCoding | 51.99% | 14 / 28 | $12.16 | |
| LiveCodeBenchCoding | 87.48% | 14 / 143 | $1.32 / $3.96 | |
| Finance AgentFinance | 57.12% | 25 / 31 | $0.82 | |
| Finance IndexFinance | 52.94% | 26 / 31 | $0.88 | |
| Tax Agent BenchFinance | 25.17% | 39 / 76 | $0.72 | |
| Legal IndexLaw | 69.01% | 15 / 28 | $0.87 | |
| Legal Research BenchLaw | 23.08% | 61 / 85 | $0.68 | |
| MedCodeMedicine | 40.45% | 65 / 107 | $1.32 / $3.96 | |
| GPQA DiamondReasoning | 89.39% | 30 / 138 | $1.32 / $3.96 | |
| MMLU-ProReasoning | 86.97% | 36 / 138 | $1.32 / $3.96 | |
| Terminal-Bench ScienceReasoning | 4.29% | 30 / 42 | $4.63 |
Vals AI, as of Oct 24, 2026.
LMArena, by category
| Leaderboard | Rating | Rank | Votes |
|---|---|---|---|
| TextLMArena | 1451 | 49 / 351 | 57.6K |
| CodingLMArena | 1469 | 70 / 346 | 17.1K |
| MathLMArena | 1455 | 68 / 339 | 2.7K |
| Creative writingLMArena | 1446 | 37 / 349 | 9.7K |
| Instruction followingLMArena | 1448 | 53 / 351 | 3.5K |
| Hard promptsLMArena | 1461 | 56 / 351 | 38.5K |
| Multi-turnLMArena | 1468 | 39 / 349 | 10.4K |
| Long queriesLMArena | 1458 | 54 / 330 | 25.9K |
| ExpertLMArena | 1463 | 68 / 305 | 5.9K |
| Legal and governmentIndustries | 1464 | 46 / 325 | 4.5K |
| Medicine and healthcareIndustries | 1469 | 38 / 325 | 3.9K |
| Business and financeIndustries | 1443 | 51 / 344 | 11.4K |
| ChineseLanguages | 1486 | 70 / 328 | 3.3K |
| SpanishLanguages | 1456 | 49 / 249 | 1.7K |
| GermanLanguages | 1458 | 42 / 261 | 997 |
| FrenchLanguages | 1474 | 41 / 252 | 2.2K |
| JapaneseLanguages | 1446 | 35 / 236 | 721 |
| WebDevLMArena | 1582 | 37 / 124 | 4.7K |
LMArena, CC BY 4.0, as of Oct 22, 2026. Where a model is rated at several effort levels, its best is shown (DeepSeek-V4-Pro, DeepSeek-V4-Pro (High, 20260813), DeepSeek-V4-Pro Preview (High)).
Compare DeepSeek-V4-Pro
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-V4-Pro?
DeepSeek-V4-Pro is made by DeepSeek, founded 2023 and based in Hangzhou.
How much does DeepSeek-V4-Pro cost?
$0.66 per million input tokens and $1.98 per million output tokens, at its provider’s list price.
What is DeepSeek-V4-Pro’s context window?
1M tokens (1,048,576).
How does DeepSeek-V4-Pro rank?
#49 of 351 on LMArena’s text leaderboard, rated 1451 (95% interval 1447–1455) from 57,588 votes, as of Oct 22, 2026.
Is DeepSeek-V4-Pro open source?
DeepSeek-V4-Pro has open weights under the MIT license, downloadable from Hugging Face (deepseek-ai/DeepSeek-V4-Pro-0813).