DeepSeek-V4-Flash vs GPT-5.6 Luna
GPT-5.6 Luna leads on 3 of 3 benchmarks where both are measured. Specs, per-token pricing, and live benchmark scores — refreshed daily from our benchmark database.
Specs & pricing
| DeepSeek-V4-Flash | GPT-5.6 Luna | |
|---|---|---|
| Company | DeepSeek | OpenAI |
| Released | Jul 2026 | Jul 2026 |
| Context window | 1.0M | 1.0M |
| Input /1M tokens | $0.098 | $1 |
| Output /1M tokens | $0.196 | $6 |
| Open weights | No | No |
Benchmarks
| Benchmark | DeepSeek-V4-Flash | GPT-5.6 Luna |
|---|---|---|
| Arena Elo | 1436 Elo | — |
| GPQA | 89.4% | 91.1% |
| Hallucination Rate (lower is better) | 95.8% | 87.7% |
| LiveCodeBench | 91.6% | — |
| MMLU-Pro | 86.2% | — |
| SWE-bench Verified | — | 93.0% |
| Speed (tok/s) | 122 tok/s | 184 tok/s |
Scores come from published leaderboards and official model cards (sources listed on the benchmarks page) and refresh daily. Pricing via vendor pages and OpenRouter.