DEEPSEEK

DeepSeek V4.1 Flash replaced V4 Flash. Official API: $0.15 input, $0.60 output off-peak. OpenRouter hosts start near $0.05. MIT weights are on Hugging Face.

DeepSeek shipped experimental vision for V4 Flash as deepseek-v4-flash-vision-exp. JPEG, PNG, GIF, and WebP; three input methods; 384 tokens per image. Here is the API contract and how to run it in OpenCode today.

DeepSeek V4 Flash routed to Novita on Vercel AI Gateway is 90% off for Pro customers through August 11, dropping the effective rate to $0.014 input / $0.028 output per million tokens. Here is the verified before/after math, the provider-pinning setup, and what a 10x cheap agent loop means for routing decisions.

The cheap coding tier repriced again: GPT-6 Luna opened at $0.10/$0.50 and took the floor from DeepSeek, whose V4.1-Flash cut rates to $0.15/$0.60 off-peak. Gemini 3.5 Flash and Claude Haiku 4.5 hold the hosted middle. Prices verified September 26, 2026.

DeepSeek re-post-trained V4 Flash into an agent workhorse: Terminal Bench 82.7, DeepSWE 54.4, native Responses API, and first-party Codex support - all at $0.14/$0.28 per million tokens. What changed, what the numbers actually mean, and how to wire it up today.

DeepSeek shipped the official V4 Flash release on July 31, 2026. The re-post-trained 0731 build beats V4-Pro-Preview on agent benchmarks at $0.14/$0.28 per million tokens. Here is what changed and how to run it through OpenCode today.

DeepSeek suspended its $74B valuation fundraising round after a leaked transcript of founder Liang Wenfeng's investor meeting laid bare the compute gap between Chinese and US AI labs - revealing he needed 200,000 Huawei 950 chips but received only 16,000.

DeepSeek, Kimi, and GLM are cheap enough to run as sidecar subagents for drafts and exploration. The catch is that cheap work you cannot inspect is just expensive noise. A shared canvas makes the output reviewable.

DeepSeek V4 Pro lands an 80.6 on SWE-bench Verified in Max reasoning mode at $0.66/$1.98 per million tokens off-peak, and Flash runs agent inner loops at $0.22/$0.66. Here is the worked cost math, the Flash-vs-Pro split, and a clear guide on when to route to DeepSeek instead of a frontier model.

A data-rich, source-cited comparison of the open-weights coding models that matter in 2026: GLM-5.2, DeepSeek V4, Qwen3, and the new Kimi K3 frontier entrant. Benchmark table, per-token pricing, context windows, self-host footprint, and a clear pick-X-if decision matrix.

deepseek-chat is deprecated and disappears July 24, 2026 - here is how to migrate to V4 Flash or Pro, with verified pricing, thinking-mode mapping, and a step-by-step checklist. Updated September 2026: the target is now deepseek-flash (V4.1-Flash).

GPT-5.4 vs Gemini 3.1 Pro vs DeepSeek V4: pricing, benchmarks, context behavior, and license terms for the mid-tier models that carry most production traffic.

DeepSeek's open-weights frontier family, previewed April 24, 2026. V4-Pro is 1.6T total / 49B active params; V4-Flash is 284B / 13B. 1M context standard. Weights on Hugging Face.

DeepSeek V4-Flash costs $0.28 per million output tokens. Fable 5 costs $50. That 178x gap is real - but so is the quality difference. Here is where it matters and where it does not.

A first-hand visit to DeepSeek HQ reveals something more interesting than benchmark scores: a 300-person company that treats AI as infrastructure, not eschatology - and what that means for API pricing everywhere.

Reasonix hit Hacker News with a DeepSeek-native pitch: keep long coding sessions cheap by designing the agent loop around prefix caching. The interesting question is when cache efficiency helps quality, and when it fights the harness.

Open-source terminal agent runtime with approval modes, rollback snapshots, MCP servers, LSP diagnostics, and a headless HTTP API. Uses DeepSeek models by default.

DeepSeek V4 is trending because it is close enough to frontier coding models at a much lower token price. The real question for developers is where cheap reasoning belongs in an agent stack.

DeepSeek V4 splits into Flash and Pro, ships a 1M context window, and undercuts every closed model on price. Here's how to wire it up with the OpenAI SDK, when to pick it over Claude or GPT, and what changed since V3 and R1.

GRPO is suddenly the standard RL recipe for reasoning models. A no-prior-knowledge mental model of PPO, GRPO, and how DeepSeek R1's training works under the hood.