Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work
DeepSeek V4 Flash is offered by 75 providers on this page. Public prices are shown for 63 of them. Its official list price is $0.15 per 1M input tokens and $0.60 per 1M output tokens. The lowest paid channel is Merge Gateway at $0.035 / $0.07 per 1M, about 15× below the list price. That channel is a third-party gateway, so confirm its availability and rate limits before depending on it. It also has 5 free ($0) channels and 5 covered by a paid subscription; free tiers usually carry rate limits, and subscription-covered access bills $0 per token only after the subscription fee.
The context window is 1,000,000 tokens at the reference host, but hosts report different limits, from 163,840 to 1,050,000, so the usable window depends on the provider you pick. It supports reasoning, tool use, and structured output. Accepted input modalities are Text and Image. The weights are open, so it can also be self-hosted or served through a gateway of your choice. Providers report the capability flags inconsistently, so verify a specific feature against the host you plan to use. Its training knowledge cuts off at 2025-05.
| Provider | Tier | Input | Output | Cache read | Cache write | Context | Output limit | Status |
|---|---|---|---|---|---|---|---|---|
| SenseNova (China) | Gateway | Free | $0 | — | 1,000,000 | 65,536 | ||
| Alibaba Token Plan | Gateway | Subscription | $0 | $0 | 1,000,000 | 384,000 | ||
| Umans AI Coding Plan umans-deepseek-v4-flash-0731 | Gateway | Subscription | $0 | $0 | 1,048,576 ⚠ | 393,215 | ||
| Alibaba Token Plan (China) | Gateway | Subscription | $0 | $0 | 1,000,000 | 384,000 | ||
| Pendra | Gateway | Free | — | — | 1,000,000 | 384,000 | ||
| UnoRouter deepseek-v4-flash:free | Gateway | Free | — | — | 1,000,000 | 384,000 | ||
| Volcengine Ark Coding Plan | Gateway | Subscription | $0 | — | 1,000,000 | 384,000 | ||
| SCNet Token Plan DeepSeek-V4-Flash | Gateway | Subscription | $0 | — | 1,000,000 | 384,000 | ||
| InferX | Gateway | Free | — | — | 1,000,000 | 100,000 | ||
| Kenari | Gateway | Free | — | — | 1,000,000 | 384,000 | ||
| DeepSeekOfficial | First-party | $0.15 | $0.60 | $0.003 | — | 1,000,000 | 393,216 | deprecated |
Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.
Interleaved thinking (reasoning between tool calls) is declared by 49 of 75 providers.
11 / 75 providers expose no reasoning control (reasoning_options: []).
10 more channels offer $0 (SenseNova (China), Alibaba Token Plan, Umans AI Coding Plan etc.); free tiers usually have rate limits and no SLA, excluded from ranking.
| Name | Conditions | Score | Metric | Source |
|---|---|---|---|---|
| SWE-Bench Verified | — | 79 | resolved | Source ↗ |
| MMLU-Pro | variant: preview checkpoint; max effort | 86.2 | EM | Source ↗ |
| SimpleQA-Verified | variant: preview checkpoint; max effort | 34.1 | pass@1 | Source ↗ |
| Chinese SimpleQA | variant: preview checkpoint; max effort | 78.9 | pass@1 | Source ↗ |
| GPQA Diamond | variant: preview checkpoint; max effort | 88.1 | pass@1 | Source ↗ |
| Humanity's Last Exam | variant: preview checkpoint; max effort; without tools | 34.8 | pass@1 | Source ↗ |
| LiveCodeBench | variant: preview checkpoint; max effort | 91.6 | pass@1 | Source ↗ |
| Codeforces | variant: preview checkpoint; max effort | 3052 | rating | Source ↗ |
| HMMT | variant: preview checkpoint; max effort · dataset: February 2026 | 94.8 | pass@1 | Source ↗ |
| IMOAnswerBench | variant: preview checkpoint; max effort | 88.4 | pass@1 | Source ↗ |
| MathArena Apex | variant: preview checkpoint; max effort | 33 | pass@1 | Source ↗ |
| MathArena Apex Shortlist | variant: preview checkpoint; max effort | 85.7 | pass@1 | Source ↗ |
| MRCR | variant: preview checkpoint; max effort · dataset: 1M context | 78.7 | MMR | Source ↗ |
| CorpusQA | variant: preview checkpoint; max effort · dataset: 1M context | 60.5 | accuracy | Source ↗ |
| Terminal-Bench | variant: preview checkpoint; max effort · v2.0 | 56.9 | accuracy | Source ↗ |
| SWE-Bench Pro | variant: preview checkpoint; max effort | 52.6 | resolved | Source ↗ |
| SWE-Bench Multilingual | variant: preview checkpoint; max effort | 73.3 | resolved | Source ↗ |
| BrowseComp | variant: preview checkpoint; max effort | 73.2 | pass@1 | Source ↗ |
| Humanity's Last Exam | variant: preview checkpoint; max effort; with tools | 45.1 | pass@1 | Source ↗ |
| MCP Atlas | variant: preview checkpoint; max effort · dataset: public | 69 | pass@1 | Source ↗ |
| GDPval-AA | variant: preview checkpoint; max effort | 1395 | Elo | Source ↗ |
| Toolathlon | variant: preview checkpoint; max effort | 47.8 | pass@1 | Source ↗ |
The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.