Leaderboard
Language ModelsOpen weights
DeepSeek V4-Flash
Efficient open-weights V4 tier (284B total / 13B active), MIT-licensed.
Crosshair Index
67.9
#15 of 32 · Language Models
- Provider
- DeepSeek
- Released
- 2026-04-24
- Parameters
- 284B
- Context
- 1M tokens
Token pricing
- Input
- $0.14 /1M
- Output
- $0.28 /1M
- Cache read
- $0.003 /1M
USD per 1M tokens · cache read = cached input (hit), cache write = caching surcharge · official list pricing (June 2026).
Industry skill web
Professional-domain strengths, composed from the benchmarks relevant to each field. Highlight an axis to see the benchmarks behind it.
Investment Banking
87
skill
Financial analysis over filings and credit agreements — valuation math, document QA, and the quantitative reasoning behind deals.
Scorecard
| Benchmark | Score | Source | Status |
|---|---|---|---|
| GPQA Diamond Reasoning | 88.1% | DeepSeek — V4-Flash model cardvendor | unverified |
| Humanity's Last Exam Frontier | 34.8% | DeepSeek — V4-Flash model cardvendor | unverified |
| SWE-bench Verified Agentic Coding | 79% | DeepSeek — V4-Flash model cardvendor | unverified |
| LiveCodeBench Coding | 91.6% | DeepSeek — V4-Flash model cardvendor | unverified |
| MMLU-Pro Knowledge | 86.2% | DeepSeek — V4-Flash model cardvendor | unverified |
| LMArena Elo Human Preference | 1,433 | LMArena (arena.ai)3rd-party | unverified |
| AA Intelligence Index Composite | 47 | Artificial Analysis3rd-party | unverified |
| Corporate Finance Finance | — | not evaluated | |
| LegalBench Law | — | not evaluated | |
| TaxEval Tax & Accounting | — | not evaluated | |
| Medical Coding Medicine | — | not evaluated |