Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding
DeepSeek V4 Flash 0731 is offered by 54 providers on this page. Public prices are shown for 50 of them. Its reference price is $0.445 per 1M input tokens and $1.34 per 1M output tokens. The lowest paid channel is Merge Gateway at $0.035 / $0.07 per 1M, about 31× below the reference price. That channel is a third-party gateway, so confirm its availability and rate limits before depending on it. It also has 1 free ($0) channel and 3 covered by a paid subscription; free tiers usually carry rate limits, and subscription-covered access bills $0 per token only after the subscription fee.
Artificial Analysis rates it 34.3 on the Intelligence Index, with 69.1 for coding and 41 for agentic tasks. Against its lowest blended price of $0.044 per 1M, that is roughly 783 index points per dollar, which is the value ratio the leaderboards rank on. Median output speed is 218 tokens per second, with 1.02s to the first token. Running one task of the Artificial Analysis suite costs about $0.2196, which reflects how many tokens its reasoning consumes rather than the unit price alone.
The context window is 1,000,000 tokens at the reference host, but hosts report different limits, from 256,000 to 1,048,576, so the usable window depends on the provider you pick. It supports reasoning, tool use, and structured output. 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.
Quality is independently evaluated by Artificial Analysis. Speed/latency are model-level medians.
| Provider | Tier | Input | Output | Cache read | Cache write | Context | Output limit | Status |
|---|---|---|---|---|---|---|---|---|
| Alibaba Token Plan | Gateway | Subscription | $0 | $0 | 1,000,000 | 384,000 | ||
| Alibaba Token Plan (China) | Gateway | Subscription | $0 | $0 | 1,000,000 | 384,000 | ||
| SCNet Token Plan DeepSeek-V4-Flash-0731 | Gateway | Subscription | $0 | — | 1,000,000 | 384,000 | ||
| Nvidia | Cloud | Free | — | — | 1,000,000 | 384,000 | deprecated | |
| Merge Gateway | Gateway | $0.035 | $0.07 | $0.007 | — | 1,048,576 ⚠ | 384,000 | |
| engy | Gateway | $0.045 | $0.09 | $0.009 | — | 1,048,576 ⚠ | 128,000 | |
| NanoGPT | Gateway | $0.05 | $0.16 | $0.013 | — | 1,000,000 | 131,072 | |
| CrofAI | Gateway | $0.08 | $0.10 | $0.003 | — | 1,000,000 | 131,072 | |
| Deep Infra deepseek-ai/DeepSeek-V4-Flash-0731 | Cloud | $0.06 | $0.18 | $0.015 | — | 1,048,576 ⚠ | 384,000 | |
| Eden AI deepinfra/deepseek-ai/DeepSeek-V4-Flash-0731 | Gateway | $0.06 | $0.18 | $0.015 | — | 1,048,576 ⚠ | 384,000 | |
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 21 of 54 providers.
2 / 54 providers expose no reasoning control (reasoning_options: []).
4 more channels offer $0 (Alibaba Token Plan, Alibaba Token Plan (China), SCNet Token Plan etc.); free tiers usually have rate limits and no SLA, excluded from ranking.
| Name | Conditions | Score | Metric | Source |
|---|---|---|---|---|
| Terminal-Bench | variant: max · v2.1 | 82.7 | pass@1 | Source ↗ |
| NL2Repo | harness: DeepSeek Harness minimal mode · variant: max effort | 54.2 | resolve rate | Source ↗ |
| CyberGym | harness: DeepSeek Harness minimal mode · variant: max effort | 76.7 | score | Source ↗ |
| DeepSWE | harness: DeepSeek Harness minimal mode · variant: max effort | 54.4 | resolve rate | Source ↗ |
| Toolathlon-Verified | harness: DeepSeek Harness minimal mode · variant: max effort | 70.3 | score | Source ↗ |
| Agents' Last Exam | harness: DeepSeek Harness minimal mode · variant: max effort | 25.2 | score | Source ↗ |
| AutomationBench | variant: max effort · dataset: public | 25.1 | success rate | Source ↗ |
| DSBench-FullStack | variant: max effort · dataset: internal | 68.7 | score | Source ↗ |
| DSBench-Hard | variant: max effort · dataset: internal | 59.6 | score | Source ↗ |
The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.
Individual evaluations run by Artificial Analysis, on the same reasoning tier as the intelligence score above. Each benchmark has its own task set and harness, so rows are not comparable with one another. The Intelligence Index above draws on a different, newer set of evaluations.