Если в проде ещё крутится deepseek-chat или вы оцениваете DeepSeek V4 GA — 20 июля 2026 выкатывается full release: peak-valley pricing, SWE-bench Verified 80.6% (open-source SOTA), 1M token context window и hard sunset legacy API 24 июля. Разбор по исходнику без воды: timeline 24.04→20.07, V4-Pro/Flash stack (CSA+HCA+mHC+Muon), benchmark table vs GPT-5.6 / Claude Fable 5, полный price sheet (CNY/USD), 6-step API migration + code samples, pain points, cost hacks и FAQ. Updated: 2026-07-20
Три месяца preview → production GA. Performance bump реальный, но commercialization + breaking API changes = operational friction. Вот что чаще всего всплывает в инженерных чатах:
Legacy countdown: deepseek-chat и deepseek-reasoner умирают 24 июля 15:59 UTC (23:59 Beijing) — немигрированный prod traffic = 404.
Peak-valley complexity: первый time-of-use pricing — будни 09–12 и 14–18 (Beijing) x2 rate; 24/7 agent pipelines ломают cost forecasting.
Closed-source всё ещё топ на hard benchmarks: Claude Fable 5 — 80.3% SWE-bench Pro; V4-Pro — 55.4%. Multi-file repo surgery = paid frontier.
Self-host hardware floor: V4-Pro (1.6T/49B active) не влезет в laptop; V4-Flash local нужен 96GB+ UMA (см. antirez ds4 + Mac 96GB benchmark).
Три inference modes: Non-think / Think High / Think Max — Think Max жрёт 384K+ context quota, cost + latency spike.
От «почти free» к regulated platform: даже после –75% в июне peak-valley делает scheduling core competency, не просто price compare.
| Дата | Событие |
|---|---|
| 2026-04-24 | V4 preview + open weights (MIT): V4-Pro (1.6T), V4-Flash (284B) |
| 2026-05 | Production-tuned V4-Flash/V4-Pro; API GA |
| 2026-06 | V4-Pro permanent –75% (output $0.87/M tokens) |
| 2026-06-29 | Email всем API users: GA mid-July + первое раскрытие peak-valley |
| 2026-07-19 | Gray-test access; медиа: «full release завтра» |
| 2026-07-20 | GA release (publish date) |
| 2026-07-24 | deepseek-chat / deepseek-reasoner permanent offline |
«Preview уже был сильным — GA = targeted bump в agent/math/code + formal peak-valley billing stack.»
DeepSeek V4 дропнул MLA из V2/V3 → CSA + HCA hybrid attention, mHC residuals, Muon optimizer. 1M context становится servable, а не marketing slide.
| Spec | V4-Pro | V4-Flash |
|---|---|---|
| Total params | 1.6 trillion (1.6T) | 284 billion (284B) |
| Active per token | 49 billion (49B) | 13 billion (13B) |
| Transformer layers | 61 | 43 |
| Context window | 1,000,000 tokens | 1,000,000 tokens |
| Max output | 384K tokens | 384K tokens |
| Precision | FP4 (experts) + FP8 (rest) | FP4 + FP8 mixed |
| Pretrain data | 33T+ tokens | 32T+ tokens |
| License | MIT | MIT |
| Mode | Поведение | Use case |
|---|---|---|
| Non-think | Без CoT, fastest TTFT | Simple Q&A, routing |
| Think High | Explicit reasoning (<redacted_thinking>) | Medium complexity, code debug |
| Think Max | Max reasoning, нужен 384K+ context | Hard math, long-chain agents |
Official sampling (all modes): temperature=1.0, top_p=1.0
V4-Pro бьёт open-weight records — closed frontier всё ещё ahead на hardest tasks. Решающая метрика: cost per task.
| Benchmark | DeepSeek V4-Pro | Claude Fable 5 | GPT-5.6 Ultra | Claude Opus 4.8 |
|---|---|---|---|---|
| SWE-bench Verified | 80.6% ★ open-source record | 96.0% | Not published separately | ~69% |
| SWE-bench Pro | 55.4% | 80.3% | 78.1% | 69.2% |
| LiveCodeBench (Pass@1) | 93.5% | 88.1% | 87.4% | 83.2% |
| Codeforces Elo | 3,206 | — | — | — |
| Terminal-Bench 2.1 | 83.9% | 88.0% | 85.1% | 82.7% |
SWE-bench Verified = real GitHub bugfixes. V4-Pro 80.6% делит open-source #1 с Gemini 3.1 Pro. SWE-bench Pro жёстче — Claude Fable 5 лидирует 80.3%.
Sources: DeepSeek docs, arXiv:2606.19348, HuggingFace, Artificial Analysis, LLMReference.
| Dimension | DeepSeek V4-Pro | GPT-5.6 Sol | Claude Fable 5 |
|---|---|---|---|
| Open source | ✅ MIT | ❌ Closed | ❌ Closed |
| Self-host | ✅ Yes | ❌ | ❌ |
| Context | 1M tokens | Not disclosed | 1M tokens |
| Code capability | Very strong (near Fable 5) | Very strong | Strongest |
| API output (off-peak) | $0.87/M tokens | ~$15/M | $50/M |
| API output (peak) | $1.74/M tokens | — | — |
| Agent orchestration | Strong, multi-agent | Strong | Strongest |
| Data privacy | ✅ Private deploy | ❌ | ❌ |
GA headline change: peak hours (Beijing weekdays 09:00–12:00 & 14:00–18:00) — rates x2.
| Model | Line item | Off-peak | Peak |
|---|---|---|---|
| V4-Pro | Input (cache hit) | ¥0.025 / $0.0035 / 1M tokens | 2× |
| Input (cache miss) | ¥3.00 / $0.435 / 1M tokens | ¥6.00 / $0.87 | |
| Output | ¥6.00 / $0.87 / 1M tokens | ¥12.00 / $1.74 | |
| V4-Flash | Input (cache hit) | ¥0.02 / $0.0028 / 1M tokens | 2× |
| Input (cache miss) | ¥1.00 / $0.14 / 1M tokens | ¥2.00 / $0.28 | |
| Output | ¥2.00 / $0.28 / 1M tokens | ¥4.00 / $0.56 |
Batch off-peak: doc processing, labeling, code review после 18:00 или до 09:00 (Beijing).
Prompt cache: static system prompts first — V4-Flash cache hit $0.0028/M.
Flash routing: simple intents → V4-Flash, hard reasoning → V4-Pro — save 60–80%.
Billing emails: DeepSeek даёт 24h notice на price changes.
Даже в peak: V4-Pro output ($1.74/M) всё ещё 8.6× дешевле Claude Opus 4.8 ($15/M) и GPT-5.6 Sol (~$15/M).
Hard deadline: deepseek-chat и deepseek-reasoner отключаются 24 июля 2026, 15:59 UTC (23:59 Beijing).
| Old model | New model | Notes |
|---|---|---|
deepseek-chat | deepseek-v4-flash (non-think) | Fast, light tasks |
deepseek-reasoner | deepseek-v4-flash (think) or deepseek-v4-pro | Pro for stronger reasoning |
Repo-wide grep: deepseek-chat, deepseek-reasoner + env vars + CI configs.
Mapping: light chat → deepseek-v4-flash; old reasoner → flash think or deepseek-v4-pro.
OpenAI SDK: swap model only; think mode via extra_body (см. code).
Anthropic SDK: base_url = https://api.deepseek.com, update model.
Staging regression: прогнать agent/RAG до 24 июля, verify peak-valley bill.
Prod canary: traffic split, rollback switch до deadline.
Think Max: context window 384K+.
# ❌ Legacy (dead after July 24)
client.chat.completions.create(
model="deepseek-chat",
messages=[...]
)
# ✅ New
client.chat.completions.create(
model="deepseek-v4-pro", # or deepseek-v4-flash
messages=[...],
# Think mode:
# extra_body={"thinking": {"type": "enabled", "budget_tokens": 8000}}
)
# Anthropic SDK path
client = anthropic.Anthropic(
api_key="your-deepseek-api-key",
base_url="https://api.deepseek.com"
)
message = client.messages.create(
model="deepseek-v4-pro",
max_tokens=4096,
messages=[{"role": "user", "content": "Hello, DeepSeek V4!"}]
)
DeepSeek V4 GA режет API bill, но local V4-Flash всё ещё требует 96GB UMA Mac; дешёвый Linux VPS не запустит xcodebuild/notarytool. Если вы крутите cloud DeepSeek API + iOS CI/CD / CLI agent long sessions — macOS heavy load на dedicated remote node стабильнее, чем bet на local iron. API pricing не чинит Keychain isolation и SSH session drops. Для prod с stable SSH, predictable bandwidth — NodeMini Mac Mini cloud rental обычно лучший execution layer независимо от peak-valley swings. Specs: цены аренды; onboarding: help center; local flash: antirez ds4 + 96GB Mac benchmark.
Updated: 2026-07-20. Закройте API migration до 24 июля.
Архитектура та же; GA bump в agent/math/code + peak-valley. Legacy models sunset 24 июля. Budget: цены аренды.
Beijing weekdays 09:00–12:00 & 14:00–18:00 (x2). Batch после 18:00, prompt cache, V4-Flash routing.
deepseek-chat → deepseek-v4-flash (non-think); deepseek-reasoner → flash think or deepseek-v4-pro. Deadline 24 июля 23:59 Beijing.
SWE-bench Verified: V4-Pro 80.6% (open-source record), Fable 5 96%. Cost/self-host: V4-Pro $0.87/M ≈ 1/17 GPT-5.6 Sol. Remote env: help center.
Да. ds4 на 96GB UMA Mac (284B/13B active); V4-Pro (1.6T) → cloud vLLM/SGLang. Details: antirez ds4 Mac benchmark.
Simple Q&A: Non-think; code debug: Think High; hard math/long agents: Think Max (384K+). Sampling: temperature=1.0, top_p=1.0.