DeepSeek V4 Flash vs GPT-5.6 Luna
DeepSeek V4 Flash and GPT-5.6 Luna both score 52 on the Intelligence Index — and both compete on price. DeepSeek wins that contest at $0.14/$0.28 per million tokens against Luna's $0.20/$1.20; Luna answers with image input, 204 tokens-per-second output, and OpenAI ecosystem compatibility. Output-heavy workloads diverge most: Luna costs four times more per output token.
Side by side
| DeepSeek V4 Flash | GPT-5.6 Luna | |
|---|---|---|
| Maker | DeepSeek | OpenAI |
| Released | 2026-07-31 | 2026-07-09 |
| Context window | 1M tokens | 1M tokens |
| Modalities in | text | text, image |
| Input $/MTok | $0.14 | $0.2 |
| Output $/MTok | $0.28 | $1.2 |
| Open weights | Yes (Open weights) | No |
Benchmark deltas
| Benchmark | DeepSeek V4 Flash | GPT-5.6 Luna | Δ |
|---|---|---|---|
| AA Intelligence Index | 52 | 52 | 0 |
Δ is DeepSeek V4 Flash minus GPT-5.6 Luna, in points. Shaded cell wins the row. Dashed values are vendor-published, not yet reproduced — methodology.
Cost calculator
- DeepSeek V4 Flash
- $9.8/mo
- GPT-5.6 Luna
- $22/mo
- Difference
- $12.2/mo
List prices as of 2026-08-11; excludes caching, batch, and long-context tiers — see model pages for tier notes.
Verdict
The budget tier now offers frontier-adjacent intelligence from two directions. DeepSeek V4 Flash is the price floor — output tokens at $0.28 per million make generation-heavy workloads (summarization, drafting, synthetic data) dramatically cheaper than anywhere else. Luna charges a modest premium for image input, faster and less verbose serving, and drop-in OpenAI API compatibility, which is often worth it purely as migration cost avoided. Text-only batch: DeepSeek without hesitation. Interactive products in OpenAI stacks: Luna.
Verdicts are analysis, not measurement, and are labeled as such per our editorial policy.
FAQ
- Is DeepSeek V4 Flash cheaper than GPT-5.6 Luna?
- Yes. DeepSeek V4 Flash costs $0.14 input / $0.28 output per million tokens against $0.2 / $1.2 for GPT-5.6 Luna (prices as of August 11, 2026).
- Which has the longer context window — DeepSeek V4 Flash or GPT-5.6 Luna?
- They match: both accept 1M tokens of context.
- Which is the more capable model, DeepSeek V4 Flash or GPT-5.6 Luna?
- They tie on the Artificial Analysis Intelligence Index: both score 52 (measured August 11, 2026). Price, speed, and modality differences decide this pairing instead.
Related news
Comparison published August 11, 2026; underlying data last changed August 11, 2026.