Migrating from deepseek-chat to deepseek-flash
What changes when you move from deepseek-chat to deepseek-flash: prices, limits, features and parameters, read from each company's own pages.
- From
- deepseek-chat
- Status
- Retired
- Shutdown
- Jul 24, 2026
- Time left
- —
- Status
- Active
- Shutdown
- none announced
- Time left
- —
Prices
USD per 1M tokens, standard tier| Type | deepseek-chat | deepseek-flash | Change |
|---|---|---|---|
| Input | Not published | $0.30first seen October 5, 2026 · source | — |
| Cached input | Not published | $0.006first seen October 5, 2026 · source | — |
| Output | Not published | $1.20first seen October 5, 2026 · source | — |
Other tiers, prompt-size bands and modalities are on each model's page.
Limits
Tokens| Limit | deepseek-chat | deepseek-flash | Change |
|---|---|---|---|
| Context window | Not published | 1,000,000 tokensin place when tracking began (October 6, 2026) · source | — |
| Max output tokens | Not published | 384,000 tokensin place when tracking began (October 6, 2026) · source | — |
Benchmarks
Same benchmark and source on both sides; settings named where they differ| Benchmark | deepseek-chat | deepseek-flash | Change |
|---|---|---|---|
| GPQA DiamondEpoch AI, max effort (vs no setting). | 71.2% | 89.8% | +18.6 pts |
| OTIS Mock AIME 2024-2025Epoch AI, max effort (vs no setting). | 48.9% | 98.3% | +49.4 pts |
A score from one source is never compared with another source's, and company-reported scores are compared only within the same company; settings named where they differ. Data: Epoch AI, Capabilities & benchmarking (epoch.ai/benchmarks), CC BY.
Features and parameters
Compare modelsNo feature is published for both models, so none can be compared.
Not published for deepseek-chat: Structured outputs, Function calling, Prompt caching, Batch API, Image input, PDF input, Reasoning / thinking, Fine-tuning, Built-in web search, temperature parameter, top_p parameter, logprobs, Context window, Max output tokens.
Not published for deepseek-flash: Prompt caching, Batch API, PDF input, Fine-tuning, Built-in web search, temperature parameter, top_p parameter, logprobs.
Aliases
All aliasesNo aliases point at deepseek-chat.
Aliases pointing at deepseek-flash: deepseek-v4-flash (in place when tracking began (October 6, 2026) · source), deepseek-v4-flash-vision-exp (in place when tracking began (October 6, 2026) · source)
On cloud platforms
All platform dates| Model | Platform | Platform ID | Shutdown | Time left |
|---|---|---|---|---|
| deepseek-flash | Microsoft Foundryversion 2026-04-23 | DeepSeek-V4-Flash | Feb 20, 2028Active | 493 days |
Read daily from each platform's own pages. The status above uses each company's own date. Time left counts to 00:00 UTC on the shutdown date, which is the evening before in ET.
Checklist
- Change the model string from
deepseek-chattodeepseek-flash.
Compared from AIWire's published data when this page loaded. Each fact links to the page it came from. Model pages: deepseek-chat, deepseek-flash.